5G Italy White Ebook

i
5G Italy White Book:

from Research to Market
Edited by:
Marco Ajmone Marsan
Nicola Blefari Melazzi
Stefano Buzzi
ii
All works of authorship contained in this publications are either original to

their respective contributors co-authors, or non copyrighted or otherwise the
object of applicable fair use provisions. In case of questions about copyright
or other intellectual property rights, please contact the publisher at:
Consorzio Nazionale Interuniversitario per le Telecomunicazioni
Viale G.P. Usberti, 181/A Pal.3
43124 Parma (PR) – ITALY
PEC: cnit@pec.it
ISBN: 9788832170016
Preface
5G is the new generation of the global telecommunication network. It is

based on a cloud-native, softwarized, end-to-end architecture, encompassing
the radio access, metro, and core network sections, as well as the edge, fog,
and cloud computing resources within the network.
Aiming to become the new reference architecture for the global mobile
and fixed telecommunication network, 5G is not only an evolution of 4G in
terms of performance, but it also creates a breaking point with respect to
previous generations: 5G will support diversified vertical services, targeting
different types of users, and including services not exclusively dedicated to
human users. Three classes of innovative service scenarios have been already
defined:
- Enhanced Mobile Broadband: that is, services characterized by very high
data rates and very high density of users, for instance virtual reality and
augmented reality, requiring extremely high quality mobile video distribu-
tion, and, in general, support of the expected increases in video consump-
tion.
- Ultra-reliable and low latency communications: that is, services related to
scenarios with demanding requirements in terms of latency and reliability,
such as automated driving, remote surgery, smart factories, smart grid,
intelligent transportation systems.
- Massive IoT (or machine type) communications: that is, the plethora of
services that will drive the advent of IoT, mainly characterized by a great
number of devices, typically requiring small volumes of data, but with
stringent constraints in terms of cost and energy.
To efficiently support end to end such diverse and demanding new appli-
cations and services, 5G is not limited to the cellular section, as previous
generations, but encompasses the whole network, and introduces several in-
novations, among which:
- New Radio, a new air interface designed to greatly improve the perfor-
mance of the access network;
iii
iv Preface
- network softwarization, i.e., the virtualization of network services, obtained

by means of three synergic solutions: i) a logically centralized control plane
enabling a flexible and agile configuration of network resources (SDN); ii)
the deployment of network functions as software components running on
industry-standard commodity hardware platforms, instead of using spe-
cialized hardware (NFV); iii) the outsourcing to computing elements (edge,
fog and cloud) not only of processing and storage functions but also of net-
work services, control and management (cloud networking);
- network slicing: a softwarized network can be virtually divided in so-called
slices, i.e., end-to-end virtual networks at the service of a class of appli-
cations; slices can be flexibly and quickly defined and setup, to support
diversified vertical services;
- full exploitation of in-network computing, following the recent trend that is
transforming the Internet in a network of data centres (or, more generally,
computing resources) in which the prevailing communication paradigm
is becoming device-to-computing-to-device, rather than device-to-device,
and in which the smartphone is becoming more and more an instrument
to access artificial intelligence services;
- a Service-Based Architecture, for which network control functions ex-
pose an Application Programming Interface (API) based on HTTP/2 and
RESTful technologies, thus harmonizing the entire network control plane
with Web technologies.
5G will constitute a larger ecosystem, including more stakeholders than
in the past, with more complex relationships, more heterogeneity and more
dynamicity. Applications sectors will be more and more actively involved
in the creation and provision of services, taking full part in the 5G value
chain. 5G will provide services not only to customers but also to industrial
stakeholders, allowing both Business to Customers (B2C) and Business to
Business (B2B) models. 5G is also an opportunity for network operators to
return in the spotlight of the service creation and management arena.
5G is very important for Italy, therefore CNIT* decided to organize on
December 4-6, 2018 the 5G-Italy event (https://www.5gitaly.eu/), which
provided a 360-degree vision about 5G. 5G Italy was a three days confer-
ence, where politics, regulatory authorities, research, businesses, economy
and public administrations met, addressing the challenges and opportuni-
ties of this technology. 5G Italy focused on policy and research issues, but
also, and significantly, on vertical applications in the most relevant sectors:
Energy, Industry 4.0, Media and Tourism, Security, Public Administration,
eHealth, Transports, Mobility and Automotive, and Infrastructure Protec-
tion. Another important aspect of the conference was the presentation of the
5G trials ongoing in 8 Italian cities.
As a companion initiative, CNIT also edited this white eBook, with the aim
of overviewing the status of 5G and of documenting the Italian involvement
in 5G research and experimentation. The eBook starts with basic definitions
Preface v
and key characteristics of this game-changing technology, proceeds to present

the work done by the most significant research and industry players, includ-
ing field trials, then focuses on applications and vertical exploitations and,
last but not least, reports on regulatory and market/business issues. A final
section of the eBook explores roadmaps and future work. Importantly, this
eBook also had a first version, printed, and presented at the 5G Italy event.
Finally, the eBook also incorporates key elements of the 5G PPP work and
results to ensure compatibility between the emerging Italian 5G ecosystem
and the pan-European 5G vision, as prepared by the 5G PPP (https://
5g-ppp.eu/).
Roma, The eBook Editors

November 2018 Marco Ajmone Marsan
Nicola Blefari Melazzi
Stefano Buzzi
vi Preface
*CNIT (National, Inter-University Consortium for Telecommunications,

https://www.cnit.it/) is a non-profit consortium established in 1995 and
bringing together 37 public Italian universities to perform research, innova-
tion and education/training activities in the field of Information and Com-
munication Technology (ICT). CNIT also owns four National Laboratories:
Multimedia Communications, Photonic Networks & Technologies; Radar &
Surveillance Systems; Smart, Sustainable and Secure Internet Technologies
and Infrastructures.
More than 1,300 professors and researchers, belonging to the member uni-
versities, collaborate within CNIT, together with more than 100 CNIT own
employees. CNIT participated in hundreds of research projects, including EU
coordinated projects, ERC grants and Italian nation-wide initiatives. In the
EU H2020 program, CNIT has obtained 41 projects and coordinated 10 of
them. CNIT has also a significant experience in the organization of scien-
tific events. CNIT’s funding comes from private companies and competitive
programs. The innovation and technology transfer of research results from
universities towards end-users and industry is a primary mission for CNIT.
CNIT also facilitates the cooperation between member universities and pro-
motes the collaboration of the same universities with other research institutes
and with national and international industries.
CNIT is very active in 5G and related initiatives: i) CNIT participates
in several EU projects on 5G and coordinates a number of them (four such
projects ranked first in their respective calls); ii) CNIT is an elected mem-
ber of the 5GPPP (https://5g-ppp.eu/), a 1.4 Billion Euro joint initiative
including the European Commission and the European ICT industries and
academia to rethink the network infrastructure and to create the next gen-
eration of communication networks and services; iii) CNIT participates in
the 5G trials of the Italian Ministry of Economic Development in Milan; iv)
CNIT participates in several EU projects on applications of 5G (e.g., for in-
telligent transport systems and autonomous vehicles) and in the Graphene
and Quantum Information flagships projects.
Acknowledgements
The Editors wish to thank all contributors, from many colleagues of Italian
academia and industry, to some very relevant international actors of the 5G
community. We are honoured to have so many high quality contributions in
this collection, which will surely further stimulate interest in 5G, and simplify
the work of many newcomers to such an exciting field.
The Editors also wish to thank Mario Alonzo, PhD student at the Uni-
versità di Cassino e del Lazio Meridionale, for his precious help with the
composition of the final book.
We hope you will enjoy reading this eBook, and possibly decide to add yet
another contribution on aspects that we may have overlooked.
vii
Foreword
Roberto Viola
DG Connect, Director General
European Commission
Our society is undergoing a rapid and profound transformation; a digital

transformation that will change our way of life forever. Yet the success of this
revolution is far from guaranteed. If it is to succeed - and I believe it must
succeed - it will be due to a combination of the right political, strategic and
economic choices. One of these is making sure that 5G connectivity is widely
available.
Without first-class communication networks, there will be no Digital Sin-
gle Market. A society built on citizens’ ability to connect to anything from
anywhere in order to manage their daily lives is only possible if connectivity
is reliable and ubiquitous. And ensuring that this required level of connectiv-
ity is there to fuel the digital revolution in Europe is one of the key tenets of
the Digital Single Market.
I firmly believe that 5G has the potential to transform our economy. It
will enable wireless broadband services to be provided at gigabit speeds, and
offer low latency and high reliability to support new types of applications,
connecting devices and objects through the Internet of Thing. It will fuel
the development of innovative business models across multiple sectors, from
transport, health and manufacturing to logistics, energy, media and enter-
tainment.
This transformation is not waiting for the advent of 5G to take place,
of course. Many of the digital changes to our society are already underway,
using existing networks. But what 5G brings them is the speed, reliability
and spread they will need to reach their full potential in the coming years.
Take AI for example: 5G will allow a significant increase in real-time data
collection and processing and distributed computing through the cloud, in
turn fuelling potential advances in artificial intelligence.
The 5G for Europe Action Plan
The potential of 5G has long been understood in the EU. We started the 5G
Infrastructure Public Private Partnership (5G PPP) - the biggest 5G research
ix
x Foreword
and development programme in the world - back in 2013. This in turn has
led to the 5G standards that we see starting to emerge today, and which are
the basis for all the large-scale 5G trials currently underway.
But getting the infrastructure and the standards in place is only part of
the picture. If we want to reach the levels of connectivity we will need for the
digital transformation of our society, we also need a long-term strategy for
how to get there.
This was the logic behind the 5G action plan adopted by the European
Commission in 2016, which aims to ensure the early deployment of 5G in
Europe. The plan set out a number of targets that will need to be reached
in order to ensure the 5G connectivity we need.
These targets include starting the first 5G trials in Europe in 2018 with a
view to full deployment in 2020, with at least one ”5G city” in every Member
State, and ensuring the availability of 5G along main transport paths in 2025.
The plan also identified the need to designate the pioneer frequency bands
to be used by 5G in order to facilitate the early trials and full commercial
introduction in 2020, as well as supporting ‘holistic’ standards not limited to
eMBB, the so-called ‘super broadband’. Finally the plan also acknowledged
the need for financial support to stimulate the growth of the market, calling
for easy access for SMEs to venture capital to catalyse the development of a
5G ecosystem.
But this is not just the position of the European Commission: in fact there
is a very wide consensus among the EU Member State governments and all the
European institutions that Europe has to act together, and quickly, to deliver
on its 5G promise. This is why we have developed together a European 5G
roadmap, a comprehensive plan for 5G development and deployment agreed
in December 2017. It is now up to each EU country to devise its own national
roadmap for 5G, and to expand their cross-border 5G trials.
The 5G Observatory – monitoring the position of

Europe in the world
It is not enough however to simply have a plan, even one agreed by every
EU Member State. We have to make sure that we stick to the plan, that we
do what we have agreed needs to be done. This is where the European 5G
Observatory comes in. Launched in 2018, the observatory monitors market
developments related to 5G and the work being done by public authorities
across the EU to prepare for the roll-out of 5G. Its first quarterly report was
published in September 2018, providing an overview of where Europe stands
in terms of 5G cities, trials and spectrum assignment, as well as the state of
play with regards to the development of national 5G strategies.
The report makes for interesting reading, in particular with regard to
Europe’s progress compared to other parts of the world. In the US, for ex-
Foreword xi
ample, at least two telecoms operators will start limited 5G-related services
in 2018, while investments in 5G infrastructure in some Asian countries, no-
tably Japan, Korea and China, are already in the billions of euros a year. But
if developments in Europe are moving at a slower pace, they are certainly
moving. In Germany, for example, Deutsche Telekom recently announced its
commitment to invest e20 billion in bringing 5G connectivity to 99% of the
population by 2025, while a number of pilots and trials are already planned
or underway ranging from 5G cities to 5G corridors for connected and auto-
mated mobility.
It is vital that Europe does not get left behind in terms of 5G develop-
ment; any delays would be detrimental not only for the telecom sector but
for the entire European economy. Since 5G is expected to drive the overall
competitiveness of European industry in the future, we need to make sure
we stick to the ambitious timetable agreed in the action plan and invest the
time, money and skills we need in ensuring comprehensive 5G coverage as
quickly as possible.
We need to learn from the lessons of the past. There were unfortunate
delays to the roll-out of 4G in several Member States which we absolutely
cannot afford to repeat. And now more than ever we are looking at these
issues within the context of the single market, which means differences in
the deployment and quality of 5G between countries must be avoided at all
costs. So we need to move quickly, but in concert to ensure full 5G coverage
everywhere.
Europe’s competitive advantage
Europe’s approach to 5G deployment is also coordinated in other ways. While

making sure that all countries act together to ensure the best possible cov-
erage is key, the European approach has also been to look at the importance
of 5G beyond the traditional telecoms sector. There is massive potential for
5G to be the basic ingredient of a whole new way of doing business, across
the whole of European industry, within the Digital Single Market. Europe
was one of the first to see this potential and to look at 5G as much more
than ‘just’ a telecoms issue; the fact that other parts of the world are coming
round to seeing it in the same way is clear validation of his approach.
It is no exaggeration to say that the innovation potential 5G brings is huge
- even if we cannot accurately predict exactly what its limits are. Who, for
example, could have predicted the rapid and global development of companies
such as Uber, AirBnB or WhatsApp on the back of 4G, which at the outset
was essentially ‘sold’ as little more a faster and more reliable telecoms service?
Yet the availability of 4G was all that was needed to spark the innovation
that led to these companies seeing the light of day.
xii Foreword
The potential, then, for 5G to drive innovation is even greater, for two
main reasons.
First, the ecosystems needed to develop the Internet of Things, where ev-
erything from household appliances satellites are potentially connected, are
already part of the 5G concept. IoT innovation is blossoming in Europe,
predicated on a reliable, comprehensive 5G network across the entire sin-
gle market, and we must ensure that we continue to fuel this innovation in
Europe.
Second, 5G will be based on cloud technologies, significantly reducing the
cost of connectivity. This is turn opens up the digital domain to many busi-
nesses in sectors beyond ICT, where the relatively high cost of connectivity
has traditionally impeded their development.
These two factors combined mean that we can expect to see the most im-
portant changes thanks to 5G not in the traditional telecoms-related sectors
but in those industries where connectivity has not been a central importance
in the past. In short, inexpensive, reliable and widespread 5G connectivity
will spark a digital revolution in anything from transport and energy to man-
ufacturing or healthcare.
And the EU is well-placed to ride this wave of innovation. The EU is
already the world market leader in many of the sectors that will likely benefit
from the increased connectivity from 5G, from the automotive sector to public
transportation or energy distribution. But the digital transformation of these
industries will not happen automatically; we need to be working already to
develop these new business and economic models to ensure that 5G is able
to meet their needs as soon as it is deployed.
The 5G Corridors for Connected and Automated

Driving
One excellent example of how this is already happening - industry and poli-
cymakers working closely together to ensure that the technology of the future
can be used already today - is the work being done on connected and auto-
mated driving applications.
Since 2007, 27 EU countries and 2 EEA countries have been cooperating on
creating of 5G cross-border corridors for large scale testing and early deploy-
ment of driverless cars. These corridors typically cover segments of motorways
of at least two neighbouring Member States to allow for uninterrupted large-
scale cross-border experimentation. Ten 5G cross border corridors will act
as test-beds for this exciting technology. Three large-scale 5G corridor trial
projects are kicking off in 2018, under the aegis of the 5G PPP. And the
Commission is already proposing to set aside EU funding to support the de-
ployment of 5G corridors within the next EU budget as part of the e3 billion
set aside for the Connecting Europe Facility Digital programme.
Foreword xiii
Conclusion
As a fully connected digital society fueled by 5G comes ever closer to becom-

ing a reality, there is plenty to be excited about. We know where we want to
go, we know how to get there. But we cannot take our foot off the pedal. We
need the continued commitment of business, of policy makers, of national au-
thorities to complete this journey. And to do this we need to make sure that
the real benefits of a 5G connected world are widely known and understood.
Publications such as this one are a vital tool in making this happen, and I
am happy to commend it to you.
Bruxelles, November 2018

Contents
Part I Introduction to 5G
1 Why 5G? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Francisco Fontes, Ioannis Neokosmidis, Riccardo Trivisonno, Franco
Davoli, Le Nguyen Binh, Spiros Mikroulis, Ioannis Tomkos, Valerio
Frascolla
2 Leading-edge 5G Research and Innovation: An undivided

commitment of Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Bernard Barani and Peter Stuckmann
3 5G as an opportunity for SMEs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
Nicola Ciulli and Jacques Magen
Part II What is 5G as of today
4 Functional Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
Andrea Detti
5 3GPP NR: the standard for 5G cellular networks . . . . . . . . . . . . . . . . 69
Michele Polese, Marco Giordani, Michele Zorzi
6 Use of millimeter wave carrier frequencies in 5G . . . . . . . . . . . . . . . . . 79
Stefano Tomasin
7 Massive MIMO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Luca Sanguinetti
8 Overview on Optical Fronthauling Technologies for Fixed-Mobile
Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Roberto Gaudino
xv
xvi Contents
9 Post-OFDM modulations for 5G and beyond . . . . . . . . . . . . . . . . . . . . 109

Paolo Banelli, Giulio Colavolpe, Luca Rugini, Alessandro Ugolini
10 Softwarization and Virtualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Giacomo Verticale, Antonio Capone
11 Efficiency/Off Grid Operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Luca Chiaraviglio, Michela Meo
12 5G security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
Giuseppe Bianchi
13 Reliable Slicing in 5G Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Luca Valcarenghi, Alessio Giorgetti, Barbara Martini, Koteswararao
Kondepu, Molka Gharbaoui, Piero Castoldi
14 5G Localization and Context-Awareness . . . . . . . . . . . . . . . . . . . . . . . . 167
Stefania Bartoletti, Andrea Conti, Davide Dardari, and Andrea Giorgetti
15 IoT Support . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Roberto Verdone, Silvia Mignardi
16 The Role of Cloud and MEC in 5G . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Fabrizio Granelli
17 5G for V2X Communications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
Antonella Molinaro and Claudia Campolo
18 Rapid Deployment of 5G Services Using Drones and other Manned
and Unmanned Aerial Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Riccardo Bassoli, Fabrizio Granelli
19 5G Planning under EMF Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . 231
Angela Sara Cacciapuoti, Luca Chiaraviglio, Gerardo Di Martino and
Marco Fiore
20 The role of propagation software tools for planning 5G wireless

networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Antonio Iodice, Daniele Riccio, Giuseppe Ruello
21 Photonics for 5G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
Antonella Bogoni, Luca Potì, Giancarlo Prati, Marco Romagnoli
22 Impulse Radio NOMA Communications in 5G and Beyond . . . . . . . . 277

Mai T. P. Le, Giuseppe Caso, Luca De Nardis, Maria-Gabriella Di
Benedetto
23 5G Trials in Italy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291
Andrea Abrardo
Contents xvii
Part III 5G research in Italy
24 Contributions to EU Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299
25 Open issues and beyond 5G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447

Marco Chiani, Enrico Paolini, Franco Callegati
Part I
Introduction to 5G
Why 5G?*
New services – Larger ecosystem – Better performance – Innovative
Architecture
Francisco Fontes, Ioannis Neokosmidis, Riccardo Trivisonno, Franco Davoli, Le

Nguyen Binh, Spiros Mikroulis, Ioannis Tomkos, Valerio Frascolla 1
Abstract The global society and the worldwide economy are becoming
increasingly dependent on information and communication technologies (ICT),
especially on wireless connectivity. The recently defined 5G infrastructure, on
the verge of being broadly deployed, promises to finally deliver on the long
sought-for ubiquitous and always-on connectivity. 5G will allow for new
applications and unique service capabilities, not only for consumers but also for
new industrial stakeholders, creating new business opportunities and allowing
for novel Business to Business to Customers (B2B2C) business models. The
society will benefit from the availability of such new services in many tangible
different ways. Nonetheless, to make such transformation happen, a larger
ecosystem, which merges existing and new ICT stakeholders, as well as
different verticals and industries, is needed. Based on an innovative architecture,
which can scale and adapt to future needs, and a new radio interface, 5G will
* A contribution of the 5G Vision and Societal Challenge Working Group of the 5G I.A., Chair,
Arturo Azcorra, IMDEA, Spain
Francisco Fontes, Altice Labs, S.A.
Ioannis Neokosmidis, inCITES Consulting S.A.R.L.
Riccardo Trivisonno, Huawei Technologies
Franco Davoli, CNIT S3ITI National Lab
Le Nguyen Binh, Huawei Research Centre, Munich
Spiros Mikroulis, Huawei Technologies Düsseldorf GmbH
Ioannis Tomkos, Athens Information Technology
Valerio Frascolla, Intel Deutschland GmbH
4 F. Fontes, I. Neokosmidis, R. Trivisonno, F. Davoli, Le Binh,
S. Mikroulis, I. Tomkos, V. Frascolla
provide significantly better performance and new capabilities, mandatory for

that transformation to take place.
1. 5G, the enabler of new services
4G was designed to improve capacity, user data-rates, spectrum usage and

latency with respect to 3G. 5G aims at being much more than a simple evolution
of 4G, as it will be a key enabler of the future digital world, the next generation
of ubiquitous, ultra-high broadband infrastructure that will support the
transformation of processes in all economic sectors and the growing consumer
market demand. This is reflected in ITU-R’s defined objective for IMT-2020:
“Enabling a seamlessly connected society in the 2020 timeframe and

beyond that brings together people along with things, data, applications,
transport systems and cities in a smart networked communications
environment”
In fact, 5G is designed to create the conditions to launch new applications

and provide new unique service capabilities not only to consumers, but also to
new stakeholders (e.g. vertical industries, novel forms of service providers,
infrastructure owners and providers).
Firstly, 5G will ensure unprecedented user experience continuity also in
challenging situations. For example, HD video will be commonplace and
available anywhere, teleworking will be possible also for those living in small
villages or travelling in high speed trains and airplanes. 5G systems will provide
user access anywhere and will select transparently for the user the best
performing access among the several available ones, based on heterogeneous
technologies like WLAN, satellite, 4G and the new radio (NR) provided by 5G.
The choice of the best performing access will not only be based on throughput
but on the most relevant metrics depending on the nature of the service; for
Why 5G? 5
instance, the right balance between latency and throughput is very important for
an Augmented Reality (AR) consumer moving around.
5G will also be a key enabler for the Internet of Things (IoT), by providing
the platform and services to connect and properly operate a massive number of
objects. To make best use of the newly provided services of 5G, the environment
in which we move will have sensors and actuators spread everywhere. Since
they require very low energy consumption to save battery lifetime, the future
network will have to find effective ways to handle a huge number of objects
requiring a highly dynamically changing amount of small energy. Objects, users
and their personal networks, whether body-worn or in a household, will be
producers and consumers of data. Future smart phones, drones, robots, wearable
devices and other smart objects will create local networks, using a multitude of
different access methods. 5G will allow all these objects to connect seamlessly
and independently of a specific access network technology.
Furthermore, several mission-critical services will be natively supported by
the 5G infrastructure, thanks to the unprecedented reliability and achievable-on-
demand low latency. 5G will cover services which were handled by specific
networks for reliability reasons such as public safety. It will also cover new
services requiring real-time reactivity, such as vehicle-to-everything (V2x)
communication services and industry applications (e.g. process automation),
paving the way towards enhanced self-driving cars, a much more advanced
factory automation or remote health services.
Figure 1: 5G new service capabilities
As a conclusion, 5G will efficiently support the three mentioned different

types of traffic profiles, namely, high throughput (e.g. for video services), low
energy (e.g. for long–lived sensors) and ultra-reliable and low latency (e.g. for
mission-critical services). In addition, the 5G infrastructure will cover the
network needs and contribute to the digitalization of vertical markets such as

automotive, banking, education, city management, energy, utilities, finance,
food and agriculture, media, government, healthcare, insurance, manufacturing,
real estate, transportation and retail.
Commercial first deployments of 5G-like systems have already started in
2018, as specifications have recently completed, defining the 5G System and a
first set of 5G services, under 3GPP Release 15, the so called 5G Phase 1.
Though a broad deployment of 5G services is expected to take place in 2020,
meanwhile the experimentation phase and the first trials are going into the field
with more and more real users, so to mature a newly defined technology and to
identify the most promising technical enhancements needed to make the 5G
proposition a reality.
2. 5G as a promoter of a new ecosystem
5G can be thought of as being more like a Business-to-Business (B2B) rather

than a Business-to-Customer (B2C) technology. By leveraging technological
advances like SDN, NFV and network slicing, 5G provides advanced
management capabilities and increased flexibility. This will enable a shift from
the traditional model of providing bundles of similar services. By slicing the
network and assigning each part a specific behaviour and capabilities, in terms
of mobility, latency, data rate, etc., 5G allows operators to offer differentiated
services using a single physical infrastructure. This in turn makes it possible to
target particular economic or industrial sectors and multiple markets. It is thus
expected that 5G will be capable of better meeting the unique needs of different
players, opening up new opportunities for both the so-called verticals and
Mobile Virtual Network Operators (MVNO).
In this new era, the traditional model of MNOs, negotiating contracts in
bilateral basis, will become obsolete as new players and roles (especially
intermediaries) will enter the value chain, thus creating new relationships
between the stakeholders, finally leading to new partnerships and innovative
ecosystems. Some of those new ecosystems have already been operating since a
Why 5G? 7
while, e.g. the 5GAA (5G Automotive Association) and the 5G-ACIA (5G
Alliance for Connected Industries and Automation), gathering under the same
umbrella entities spanning very different types of activity (e.g. operators, car
manufacturers, industries, telecom equipment vendors, software houses, research
institutes and consumers). Moreover, the need of network densification will
create new opportunities for “upstream” players like independent operators that
will acquire sites, deploy their infrastructure and offer wholesale to operators.
This model, usually known as neutral host, will be mainly applicable in dense
urban areas and indoor public spaces, and will allow partnerships and co-
investment schemes between operators, vendors and verticals.
The ongoing discussions in standards on the next features to be added to the

initial 5G deployment hint at the concrete possibility to finally deploy local
private networks, thus allowing for unprecedented use cases and services, of
which several different verticals will benefit, like media and entertainment or
factory automation, among others. Additionally, compared to previous
generation systems, 5G standards are including significant enhanced capabilities
allowing 3rd parties (e.g. verticals) to customise network architecture and
policies.
Last but not least, owners of sites suitable for 5G antenna installations, such
as landlords or stadium owners, will gain significant negotiating power as the
supply of available sites will be lagging behind the demand for such sites.
Infrastructure resources, connectivity and all network functions will be delivered
as a service. The ensuing environment will foster partnership-based business
models. Operators will tap into the opportunity to enhance the value of third-
party services. Partnerships will be established on multiple layers ranging from
sharing the infrastructure, to exposing network capabilities as a service end-to-
end, and integrating partners’ services into the 5G system, through a rich and
software-oriented capability set.
With 5G, network services will rely massively on software, with the cloud
computing model extending to the telecom industry and cloud-native
applications evolving to 5G-ready applications, with a stronger awareness of
networking capabilities. Larger IT providers have already penetrated this market
recently, relying on their expertise in cloud computing and virtualization to
provide the same value proposition towards the telecom sector. It may cause a
disruptive impact to network manufacturers which will reposition themselves,
with a rollout of software solutions from their in-house development labs, or

strategic partnerships with new IT providers.
Edge clouds will emerge at locations only to be made accessible by
operators, as mandatory for 5G to guarantee the promised low latency. Besides
services and applications running over-the-top (OTT) and other interacting
directly with 5G core, a new generation of providers will take benefit of this new
location, building a new ecosystem right at the edge. We will see some specific
network platforms for each vertical sector with dedicated features and
performance requirements (e.g. high reliability for health or automobile verticals
or high density of terminals for smart cities).
Vertical industries will benefit from this environment, by being enabled to
build their applications on top of a flexible network, capable of providing
networking services through network slices tailored to a specific vertical and to
adapt their configuration accordingly to the vertical application’s evolution.
These capabilities will create further opportunities for business interactions
among multiple network service providers, vertical industries and application
developers, fostering a dynamic market evolution, which will be instrumental to
create new working opportunities, thus finally impacting the whole society in a
positive way.
3. 5G enhanced performance
5G is mandated to significantly increase the available communication system

performance, according to several system parameters and Key Performance
Indicators (KPI). As recommended by ITU-R, 5G is therefore not only expected
to be disruptive but also to act as an economy booster by fostering the creation
of (i) new services that will positively impact important societal aspects, (ii) new
ways to seize business opportunities, especially for the new service providers, as
well as (iii) new business models supported by advanced ICT technology
enablers.
The 5G architecture and its underpinning technologies will allow the usage
of network functions and resources that are tailored to optimize specific
Why 5G? 9
services. Furthermore, allowing for even more advanced sharing infrastructure

and spectrum capabilities, 5G will help reducing deployment costs and foster the
entrance in the market of new players, thus enriching and enlarging the
ecosystem of service providers.
Those disruptive capabilities will provide ubiquitous access to a wide range
of applications and services, and will allow for increased resilience, continuity,
much higher resource efficiency, and an overall significant decrease of system
energy consumption. At the same time, 5G will increase the level of security and
privacy of future communications. In addition, 5G will provide enormous
improvements in capacity and boost user data rates. For instance, peak data rates
up to 20 Gb/s will be available. A capacity of 10 Gb/s/km² will be required to
cover, e.g., a stadium with 30,000 devices relaying the event in social networks
at 50 Mb/s. Moreover, reduced end-to-end latencies of the order of a millisecond
are needed to support immersive interactive applications and ensure ultra-
responsive mobile cloud-services.
Besides the human-centric applications, some of which have been outlined
above, it is expected that a wide variety of Internet of Things (IoT), massive
Machine-Type Communication (mMTC) and especially Ultra-reliable Low-
Latency Communications (URLCC) applications will be mainstream by 2025.
The capability of the telecommunication system to fulfil the numerous and
diverse new requirements coming from the above-mentioned applications and
the 5G system verticals will require important changes to the currently defined
and implemented architecture components of mobile systems.
Figure 2, taken from the ITU-R [1], document “enhancement of key
capabilities”, compares existing and forthcoming system parameters and KPIs
and highlights the main benefits expected from 5G.
Figure 2: IMT-2020, enhancement of key capabilities [1]

It is important to highlight that not all of the enhanced capabilities described

above will be required by each 5G service, everywhere and all the time. Each
connected device will typically have its mix of latency, bandwidth and traffic
intensity requirements. Also, each connected area will have its specific
characteristics: the network will not provide the same coverage for a business
district, a stadium, a residential area, or on board a vehicle. This is why the
infrastructure has to be enhanced so to be flexible and able to dynamically adapt
itself to the characteristics of the specific service demand expected in that
particular area. 5G is therefore expected to provide, for instance, extended
coverage for mMTC services and high bandwidth for eMBB services.
Radio based services rely on regulated access to electromagnetic spectrum,
only at specific frequencies. In order to provide very high overall system
capacity, so to fulfil the new 5G service requirements, especially the eMBB
ones, it is required to make use of very wide contiguous carrier bandwidths,
from hundreds of MHz up to several GHz. That will be possible, if contiguous
frequencies are to be used, only thanks to higher carrier frequencies, i.e. well
above 6 GHz, in the lower millimetre wave spectrum.
4. Innovation, at the heart of 5G
To achieve the expected enhancements in capacity, coverage, reliability, latency,

energy consumption, flexibility, availability and fast services deployment, a new
system architecture and a new radio interface are at the core of the 5G
proposition. However, besides the intrinsic innovations introduced by the
technology, the 5G system architecture and its deployment will leverage on
other emerging, innovative, technologies, its success being the result of all those
combined together.
A new 5G Core to interconnect everything

Why 5G? 11
The architectural changes to the telecommunication systems, needed to fulfil the

expected 5G business and performance requirements, are significant compared
to the existing deployed networks. In order to realize such radical changes, the
various 5G subsystems and interfaces, as well as their integration into the overall
5G substrate and the interwork with the legacy systems, need to be inspired by
modern operating system architectures.
The design principles of 5G Core service based network architecture (defined in
the latest standard specifications [2]) can be synthesized in the following points:
1. Modularized network architecture, defining a plurality of elementary
network functions enabling flexible and efficient network slicing
2. Service Based Architecture (SBA), allowing direct interaction among
network functions, via the exposed services
3. Minimization of dependencies between the Access Network (AN) and the
Core Network (CN)
4. Support of "stateless" Network Functions, where the "compute" resource is
decoupled from the "storage" resource
5. Support of concurrent access to local and centralized services, to support
low latency services and access to local data networks
6. Introduction of data analytics capabilities, to enable enhanced network
automation
7. User Plane (UP) and Control Plane (CP) separation, as already introduced in
previous generation networks.
These characteristics allow 5G deployments to adapt to a wide range of

scenarios and appropriately exploit the adoption of technologies like SDN, NFV
and MEC for 5G deployment, operation and services implementation. End to
end slicing will be a central feature and selling argument of 5G, with explicit,
native support, which exploits virtualization and much of the characteristics
presented above.
A new radio interface to connect everything
One of the key enhancements of 5G is its new radio interface, called 5G New
Radio (NR). Designing a wireless access network that simultaneously satisfies
future demands for both human-centric and machine-centric services calls for
technologies that present some characteristics:
1. Operation from low to very high bands: 0.4 – 100Ghz, including standalone
operation in unlicensed bands
2. Up to 400 MHz component-carrier bandwidth (up to 100MHz in <6GHz; up
to 400MHz in >6GHz) and up to 16 component carriers
3. Set of different numerologies for optimal operation in different frequency
ranges
4. New channel coding (LDPC for data channel, Polar coding for control
channel)
5. Native support for Low Latency (shortened Transmission Time Interval
(TTI))
6. Native support for ultra reliability (Multiple diversity mechanisms)
7. Flexible and modular RAN architecture: split fronthaul, split control- and
user-plane
8. Support for devices connecting directly, with no network (D2D, V2x)
9. Native support for Network Slicing
5G network deployments are expected both in the “low” bands (i.e. in
frequencies below 6GHz) and in the “high” bands (i.e. in frequencies above
6GHz), using macro and small cells, coexisting with cellular legacy technologies
(2G, 3G, 4G/LTE) and other wireless technologies like WiFi.
In certain areas (especially urban and indoor) and to guarantee the required
coverage and performance, 5G deployment will be based on Ultra-Dense
Networks (UDN), with the deployment of a significant number of small cells,
coexisting with macro cells. This will put pressure on costs and the transport
network, as presented below.
A wireless network made of optics
With the advent of the 5G era, requirements like CapEx reduction,

performance and energy efficiency (OpEx) improvements, capacity sharing and
optimization, together with the need to deploy dense radio networks, are
emphasized. As a consequence, the need to have the Radio Access Network
(RAN) evolving from monolithic deployments to a generalized Cloud RAN (C-
RAN), where Radio Units (RU) units are connected to Distributed and
Why 5G? 13
Centralized Units (DU/CU), in a balanced splitting of functions, emerges as

being fundamental. Thus, appropriate RAN functions split, interconnected by
appropriate front/mid-haul optical fiber, packet based solutions, will be key in
the 5G RAN architecture, answering the strict requirements of the 5G enablers,
such as massive MIMO, mmWave and ultra-dense cells.
5G as well as data centres will be key technological and business growth
drivers for the optical communication industry. The Optical Transport Network
(OTN/WDM) is a robust solution which can provide a reliable broadband
fronthaul delivery, using high capacity DWDM and Ethernet (like IP/Eth-
eCPRI), being a promising option to support functional splits. Passive Optical
Networks (PON) can reduce the cost of fiber deployment and reuse the current
optical distribution network, subject to bandwidth upgrade (e.g. via NG-PON2),
low latency and protection-multi-wavelength allocation requirements. Radio
over Fiber (RoF) can be considered in the long term, as Digital RoF (CPRI like)
cannot cope with mm-wave technology.
Furthermore, breakthrough technologies in optical interfaces and
components will be needed. Especially advanced technologies like Silicon
Photonics can be the cost-efficiency enabler for C-RAN, based on optical
fronthaul and Wavelength-Division Multiplexing (WDM).
Software based architecture
5G will be driven by software. Network functions are expected in a number

of points of presence, from the core to the edge of the network, to meet
performance targets. As a result, 5G will heavily rely on emerging technologies
such as SDN, NFV and Multi-access Edge Computing (MEC) to achieve the
required performance, scalability and agility.
The search for solutions to make resource allocation in telecommunication
networks more dynamic, performance-optimized and cost-effective, has brought
forth the characterizing three features of flexibility, programmability and
energy-efficiency. The first two are addressed by Software Defined Networking
(SDN) and Network Functions Virtualisation (NFV). In particular, in the vision
of ETSI [3], the third one leverages “…standard IT virtualisation technology to
consolidate many network equipment types onto industry standard high volume
servers, switches and storage”, which could be located in Datacentres, Network
Nodes and in the end user premises. The 5G architectural framework is

expected to instantiate on these, achieving unprecedented flexibility,
performance and efficiency, integrating mobile wireless and fixed network
components in a unified heterogeneous networking framework.
Complementarily, NFV and SDN allow more flexibility and tighter
integration with infrastructure layers. However, none of the two is essentially a
networking technology, as both assume an underlying network on top of which
to operate. Hence, 5G will provide a unified control for multi-tenant networks
and services through functional architectures deployment across many
operators’ frameworks, giving service providers, and ultimately prosumers, the
perception of a convergence across many underlying wireless, optical, network
and media technologies. 5G will make possible the fundamental shift in
paradigm from the current “service provisioning through controlled ownership
of infrastructures” to a “unified control framework through virtualization and
programmability of multi-tenant networks and services”, natively supporting and
exploiting slicing.
Artificial Intelligence powering 5G
While the diversity of services and the complexity of the infrastructure will
increase with the introduction of 5G, 5G is also expected to radically cut
infrastructure Total Cost of Ownership (TCO), as well as the service creation
and deployment times. Hence, service and network management, which
classically rely on the Operation Administration and Management (OA&M)
tools and the Business and Operations Support Systems (BSS and OSS), will
evolve accordingly with advanced automation, including cognitive operations
for handling trillions of actuators, sensors, and exploiting Big Data for better
QoS and QoE, whatever the prosumer will be (human, machine or thing).
Considering the expected significant dependency industry and society will
have on 5G, the development of cognitive features, as well as the advanced
automation of operation through proper algorithms, will be paramount to
achieve the required availability and reliability levels, and to optimize complex
business objectives, such as end-to-end energy consumption. In addition, the
exploitation of Data Analytics and Big Data techniques, via the incorporation of
Why 5G? 15
Artificial Intelligence, will pave the way to monitor the users’ Quality of
Experience through new metrics combining network and behavioural data, while
guaranteeing privacy.
5. 5G Standardization and launch of first commercial

services
Several standards bodies are involved in the definition of the forthcoming 5G

system. The most important one is the 3rd Generation Partnership Project
(3GPP), even though others, like e.g., ETSI and IEEE, are also working on
defining some aspect of the 5G system.
In September 2018, 3GPP has completed the definition of the first set of 5G
features, called 5G Phase 1, under the stream of activities of the so-called 3GPP
Release 15. This first set defines the new telecommunication architecture
components and protocols and messages needed to make interwork the 5G new
access technology, called 5G New Radio (NR), with the legacy systems. From
the usage scenarios point of view, 5G Phase 1 has focused on mainly defining
the eMBB services, whereas the other two main usage scenarios, i.e. URLLC
and mMTC will be the focus of the ongoing 5G Phase 2, under the activities of
the 3GPP Release 16, planned to be completed by the end of 2019. The content
of what will come afterwards, most probably called 5G Long Term Evolution
(5G LTE), under the work of 3GPP Release 17, is still a matter of speculation.
ETSI and IEEE are also working on some aspects of 5G system. For
example, the IEEE 1932-1 group on licensed/unlicensed spectrum
interoperability and ETIS RRS groups on eLSA, an extension of the LSA
(Licensed Shared Access) concept, needed to fulfill the requirements of 5G
verticals, like Media and entertainment.
2018 is also the year of the first commercial launch of 5G services. For
instance, since October, Verizon in US leverages the capability of 5G
connections to provide a better user experience in households. More launches
are planned at the beginning of 2019, but it is in 2020 that a real broad
deployment of 5G services is expected to take place. Meanwhile, the
experimentation phase and the first trials are going into the field with more and
more real users, so to make mature the newly defined 5G technology. Such trials
and first commercial launches are key to identify the most promising technical
enhancement needed to make the 5G system proposition not only a reality, but
also a commercial success.
References
1. ITU-R, ITU-R, Rec. M.2083, "IMT Vision – Framework and overall objectives of the future
development of IMT for 2020 and beyond", 2015.
2. 3GPP, TS 23.501, "System Architecture for the 5G System", Stage 2, Release 15.
3. ETSI, “Network Functions Virtualisation (NFV)”, White Paper, 2013
Leading-edge 5G Research and Innovation: An
undivided commitment of Europe1
Bernard Barani and Peter Stuckmann2
Abstract Research and Innovation paving the way towards the development of
5G Communication Networks has been subject of huge support and political
commitment from Europe, especially under the Horizon 2020 programme. The
European Commission has notably implemented with industry the 5G Public
Private Partnership as an R&I vehicle to structure and foster European research
in this domain and also to further support the deployment agenda set out in the 5G
Action Plan. This paper reviews the main development and impacts of the 5G PPP
R&I actions and outlines future actions.
1
NB: The views expressed in this article are those of the authors and shall not be
considered as official statements of the European Commission.
Bernard Barani
European Commission, DG CONNECT-E1
Peter Stuckmann
European Commission, DG CONNECT-E1
18 Bernard Barani and Peter Stuckmann
1 Introduction
Early reflection about the evolution of mobile communication networks “beyond
4G” started soon after the first deployment of a 4G commercial network in
Sweden, in 2010. In those days, it was already apparent that the very fast growth
of mobile traffic, between 50 to 100% increase on a yearly basis, as well as the
prospects to serve innovative Internet of Things (IoT) applications would drive
further R&D in the mobile communication domain.
Taking note of these developments, the European Commission initiated visionary
EU-funded research activities already in 2012i. At the Mobile World Congress in
2013, Commissioner Kroes challenged the industry to come up with a structuring
European approach for leading edge R&D in 5G network technologies and
systems. This eventually led to the setup of the European 5G Public Private
Partnership (5G PPP). The 5G PPP is implemented under the European Horizon
2020 programme with about € 700 Million of public support over the 2014-2020
period. The private sector contribution is matching that amount by a factor of at
least five. Altogether, this represents the largest 5G R&D initiative in the world.
Piggybacking on these intense technological efforts, and taking stock of fast
international developments, Commissioner Oettinger made a formal call to the
European industry at the Mobile World Congress in 2016 in view of developing
an ambitious 5G deployment roadmap for Europe. Industry responded with a 5G
manifestoii and the Commission adopted the 5G Action Plan (5G AP) on 14
September 2016iii as part of a comprehensive connectivity package setting out the
European ambitions for a Gigabit Society.
These initiatives materialise the importance of 5G networks for Europe. They are
considered by the European Commission as a strategic asset for the digital society
and to support the digital transformation of the industry and the public sector.
2 5G Vision driving R&D and technological

requirements
There are multiple socio-economic developments driving the telecom and the
wider ICT sector. Broadband access has become the norm and the advent of ever
Leading-edge 5G Research and 19
Innovation: An undivided
commitment of Europe
more feature rich content located in remote clouds coupled with ever more
powerful end user devices like tablets and smartphones call for networks of ever-
higher capacity and speeds. Bandwidth consumption of mobile networks, even if
one order of magnitude lower than on fixed networks, continue to grow at a rate
of at least 50% in most countries, mainly due to video traffic. The advent of novel
bandwidth hungry Virtual or Augmented reality (VR/AR) mobile applications
will further exacerbate this trend. In addition, the advent of the Internet of things
(IoT), with massive deployment of connected objects in cities or in dense location
areas calls for new approaches to efficiently address huge collections of devices
with minimum power consumption and efficient connectivity. Finally, the advent
of new mission critical applications where response time is of the essence, such
as in factories environments, healthcare, public protection or automated driving
calls for extremely low latency systems with very high availability and reliability
characteristics, beyond what 4G is capable of delivering. In fact, 4G design drivers
were mainly based on mass-market access to high speed mobile Internet, whilst
5G also takes into account applications in professional environments requiring
much higher performance and grade of service levels.
These novel requirements for future 5G networks were further refined by
industry in several documentsiv, notably at ITU level. They cover: i) the "enhanced
Mobile Broadband” (eMBB) scenario targeting carrier data rates larger than 10
Gb/s; ii) the massive Machine to Machine communication scenario (mMTC)
targeting connectivity of millions of devices per km²; iii) the Ultra Reliable Low
Latency Communications (URLLC) scenarios, targeting latencies in the order of
1ms at the level of the User Plane. The main resulting radio requirements as
worked out at ITU level are outlined in the table below v.
IMT-ADVANCED IMT-2020 (5G)
DL: 1 Gbps DL: 20 Gbps

PEAK DATA RATE
UL: 0,05 Gbps UL: 10 Gbps
USER
EXPERIENCED 10 Mbps 100 Mbps
RATE
PEAK SPECTRAL DL: 15 bps/Hz DL: 30 bps/Hz

EFFICIENCY UL: 6,75 bps/Hz UL: 15 bps/Hz
MOBILITY 350 km/h 500 km/h
USER PLANE
>10 ms 1 ms
LATENCY
CONNECTION
x1000 devices/km² 1 million devices/km²
DENSITY
NETWORK
ENERGY 1 – Normalised x100 over IMT-Advanced
EFFICIENCY
AREA TRAFFIC
0,1 Mbps/m² 10 Mbps/m² (hot spots)
CAPACITY
BANDWIDTH UP TO 20 MHz/channel Up to 1 GHz/channel
Table 1: 5G main radio KPI’s
Based on this early vision, the 5G PPP developed further an "EU vision for 5G”,
where vertical use cases are key drivers for 5G developments. This was outlined
in a White Papervi released at the Mobile World Congress 2016. It describes a
European approach with 5G called upon to implement a more holistic and radical
network transformation to serve vertical industries, with connectivity solutions
tailored "ad-hoc" to the specific digital business case of diverse industries (e.g.
automotive, health care, smart factories, energy, media). This vision takes
advantage of the introduction of technologies inspired from the IT/cloud
computing domains such as Network Function Virtualisation (NFV) and Software
Defined Networks (SDN) notably used to realise network slices over multiple
domains and tailored ad-hoc to the various application requirements of multiple
tenants. In this approach, the role of connectivity also shifts from a "cost factor"
to an intrinsic asset of a full digital product or service. This strategy relies on the
development of cross sectors ecosystems, beyond the provision of shorter-term
super high rate access. It is directly in line with the wider policy ambitions of
"Digitisation of the European Industry" vii (DEI) as presented by the Commission
in April 2016.
Connected and Automated Mobility (CAM) as a lead “5G vertical"
In the context of 5G serving a multiplicity of vertical sectors, the deployment of

5G infrastructure along main transport paths in Europe by 2025 is one of the three
strategic connectivity objectives set out by the Commission in its 5G Action Plan.
The expectation is that 5G connectivity will be a major enabler for Connected and
Automated Mobility, a key opportunity for Europe to lead in digital innovation.
All along the main pan-European transport paths, vehicles should be able to move
across borders with uninterrupted 5G connectivity and guaranteed quality of
service level to ensure business continuity for CAM applications, i.e. while
changing operational, regulatory and administrative environments.
Against this background, the Commission is encouraging cooperation between
Member States on cross-border initiatives for the establishment of large scale
testing and early deployment of 5G corridors, including on aspects related to the
cross-border exchange of road safety and traffic information, data access, data
quality and liability.
In 2017 and 2018, several Member States and EEA (European Economic Area)
countries signed Letters of Intent (LoI) to establish 5G cross-border corridors for
large scale testing and early deployment purpose. Corridors typically cover
segments of motorways of at least two different neighbour Member States
allowing for uninterrupted large-scale cross-border experimentation or early use
of 5G for CAM.
At this stage, ten such corridors are available across neighbouring Member States,
as outlined in the table below. These corridors are open for implementation of
pilot experimentations of cross border CAM systems and services using the most
advanced 5G technological capabilities.
Metz-Merzig-Luxembourg: FR-DE-LU
Rotterdam-Antwerpen-Eindhoven: NL-BE
Porto-Vigo, Evora-Merida: PT-ES
E8 "Aurora Borealis": NO-FI
Nordic Way2: NO-SE-FI-DK
Brenner Corridor: IT-AT-DE

Thessaloniki, Sofia-Belgrade: EL-BG-RS
EE-LV-LT Via Baltica (E67) Tallinn (EE) – Riga (LV) – Kaunas (LT) –
Lithuanian/Polish border
LT-PL Via Baltica Kaunas-Warsaw
Greece-Turkey (8 km segment across the border)

Table 2: List of currently available 5G cross border corridors
3 Economic Opportunities
From a market perspective, 5G revenues may reach US$250 billion in 2025 with
North America, Asia-Pacific, and Western Europe being the top markets viii, of
which critical and massive Machine-to-Machine communications will potentially
generate significant revenues in addition to enhanced Mobile Broadband services.
A study carried out for the European Commissionix indicates that the full benefits
of the future 5G capabilities in Europe over 4 industrial sectors (automotive,
healthcare, transport, utilities) may reach €113 billion per annum on the long run.
In the year 2025, € 62.5 billion could already arise from the first order benefits in
these four key industrial sectors. The same study also concludes that 5G
introduction in Europe has the potential to generate 2 million jobs.
Other studiesx led with a global perspective indicate that 5G penetration in 8
different industrial sectors would generate a 34% growth of the connectivity
business in 2026, adding more than € 500 billion globally to the classical
broadband revenues whose growth is expected to be much smaller.
Lead industry actors also predict that 5G will already represent more than 550
million connections in 2022xi globally, more than the current 150 million LTE
subscriptions in Europe.
Altogether, the market prospects offer significant economic opportunities whilst
expected saving and efficiency gains in vertical industries will also contribute to
societal goals such as decreased environmental footprint or decreased number of

road fatalities.
4 Technologies to realise the 5G vision

The implementation of the EU vision of 5G set out by the 5G PPP addresses a
wide range of technologies. The Commission has recognised the need to move
towards the "Gigabit society"xii. Meeting this objective can be greatly facilitated
by the increased use of wireless technologies such as in the context of 5G, taking
as a target the ITU objectivesxiii of moving beyond 10 Gb/s on the radio access.
Today, the maximum throughput envisaged with LTE-A-PRO is in the order of
3.2 Gb/s, with carrier aggregation across several bands. Carrier aggregation has
eventually some limitations, considering the growing complexity of devices
integrating several bands and the fact that multi-band combinations may come at
a loss of 20% or so of spectrum efficiency. This has prompted industry to consider
the use of higher frequency bands at millimetre-wave frequency ranges, where
large chunks of contiguous spectrum are available. Over the last few years, several
industrial trials have demonstrated the transmission capability of higher frequency
bands, (e.g. 15, 28, 73 GHz) to support data rates above 10 Gb/s, either in fixed
or mobility conditions.
Still, the actual delivery of 5G capabilities requires going much beyond the
availability of a new high-speed radio interface. Multiple technologies are called
upon to achieve at least:
- a flexible radio access network that allows operators to manage an heterogeneous
set of access technologies and to optimise the access according to the required
service needs and to manage multiple radio accesses as a seamless access
continuum across multiple frequency bands ranging from UHF to millimetre
waves. This is needed to address the wide range of application requirements
targeted by 5G, taking into account that different radio accesses at different
frequency bands exhibit different coverage, bandwidth and grade of service
characteristics;
- a large range of deployment scenarios, including a variety of static or moving

nodes, with much denser deployment of access points, integrated
backhaul/fronthaul operations, and optimised locations of Centralised Units (CU)
and Distributed Units (DU) in the context of Cloud-RAN (C-RAN)
implementations;
- very low latency services, with optimisation at several levels, e.g. at air interface
level with MAC design enabling fast access and low Transmission Time Interval
(TTI), and at architectural level using Mobile Edge Computing and in network
caching techniques;
- massive connectivity services, with redesign of access protocols enabling to
drastically reduce the signalling load over the air interface, whose overhead tend
to grow very fast as large amounts of devices with small bursty traffic try to access
a resource pool;
- high performance in high mobility scenarios, with control of Doppler effects at
higher frequency ranges, use of MIMO techniques and optimisation of handover
overhead in high density deployments.
The above issues may be considered as a non-exhaustive list of issues driving the
industrial research agenda globally. A White Paper, presented by the 5G PPP in
the context of the Mobile World Congress 2017xiv details the contribution of
European R&D to these important issues.
Beyond these aspects, mostly related to Radio Access Network (RAN)
architectures and technologies, the full transformative value of 5G requires the
adoption of NFV and SDN technologies on a large scale to support a redesigned
core network. This will be necessary to make 5G a truly holistic orchestration
platform that integrates networking, computing and storage resources into one
programmable and unified infrastructure. It embodies the vision of a flexible
multi-tenant architecture where computing resources are distributed within the
network including operational sites of the vertical industry stakeholders, within
the base stations, in edge clouds at central offices, in regional and central clouds,
and managed by different stakeholders.
The full realisation of 5G hence calls for a "next generation" Core Network
architecture, based on SDN/NFV paradigms, to address an Access Agnostic
Converged Core Network, enabling next generation services regardless of access
network and integrating next generation devices. It feature fully flexible,

programmable separate Control and Data planes, unified connectivity, security,
mobility and routing management, as well unified physical infrastructure and
corresponding abstractions (virtual resources, functions, etc.) control and
orchestration.
Eventually a Network Operating System may be called upon to manage and
orchestrate a unified access to computing, storage, memory and networking
resources across wired networks, wireless (cellular and satellite, access and
fronthaul/backhaul) networks. This requires the identification of abstractions of
primitives, functions and corresponding states, in the control and data planes for
a unified connection, security, mobility and routing management. These aspects
are currently subject of intensive research work worldwide. The 5G PPP has
released an Architecture White Paper addressing these issues xv
5 The 5G Public Private Partnership (5G PPP)

The implementation of Research and Innovation actions under the 5G PPP has
been driven by a 5G roadmapxvi developed in 2013 and forming the basis of the
contractual agreement signed between the European Commission and the private
side, represented by the 5G Industrial Association (5G-IA). The 5G-IA currently
include 52 membersxvii, from industry, research centres, academic institutions
and user companies. SME’s are also well represented, as one of the objective of
the 5G PPP is to involve at least 20% of SME’s as beneficiaries of the funded
actions.
The definition of the R&I agenda as well as the updates of the roadmap are
managed by the 5G-IA and supported by the NetWorld2020 European
Technology Platformxviii (ETP) a body representing more than 1000 research
organisations in Europe. This ensures an inclusive, open and transparent process.
The 5G PPP roadmap has been defined with three phases of collaborative
research:
• Phase 1 addressing fundamental research on key technologies and
architectures needed to support the 5G Vision.
• Phase 2 addresses integration of core technologies towards the development

of Prof of Concepts (PoC’s), prototypes and trial involving technology
validation in the context of a multiplicity of vertical use cases;
• Phase 3 moves forward European trials and pilots, by providing a pan
European end-to-end 5G experimental platform supporting the
implementation of vertical large-scale trials in a multiplicity of sector. In that
context, CAM receives particular attention. Phase 3 directly supports the
deployment objectives set out in the 5G Action Plan.
At this stage, 49 projects have been implemented under the 5G PPP. They
include 433 different organisation and include thousands of researchers and
developers across Europe which have been working on innovative solutions for
the definition and use of 5G. The distribution of efforts in from the participating
nations is illustrated in the figure below.
5G PPP budget distribution
DE ES FR UK IT EL SE FI IL PT NL IE
BE NO DK LU PL CY TR AT RO HU GI SI
Figure 1.
Contrary to classical R&I implementation, contracted projects do not run in
isolation. A key part of the 5G PPP structure is a set of cross-projects and cross-
initiative working groups. The working groups are the means to establish and
publish program level positions on issues that impact global 5G developments or

may be the basis for liaison or interaction with external bodies such as other
regions or standards bodies. Main deliverables of this joint work include:
• 5G PPP 5G Architecture White Paper Revision 2.0xix (December 2017)
Highlighting the key 5G architecture design recommendations from 5G PPP
Phase 1 and providing a baseline architecture for Phases 2 and 3.
• 5G PPP Security Landscape (June 2017)xx
Providing insights into how 5G security should be addressed in terms of
“what” and “why”.
• 5G Innovations for new Business Opportunities (March 2017)xxi
• Showing how the 5G PPP innovations go beyond early 5G deployments for
the eMBB service class, and how all 5G service classes may be delivered
over a scalable and cost-efficient network. It explains how 5G technological
innovations transform the network into a secure, reliable and flexible
orchestration platform across multiple technology and administrative
domains.
• 5G PPP Cognitive Network Management for 5G (March 2017)xxii
Presenting the novelties for network management in 5G.
• 5G PPP Vision on Software Networks (January 2017)xxiii
Providing a first conceptual architecture seamlessly and flexibly combining
SDN and NFV technologies for 5G.
Such collaborative work is notably leveraged to create industrial consensus and
to support standardization work. At this stage, it is estimated that 5G PPP projects
running under phase 1 have been at the origin of more than 320 industry
contributions to standardization bodies and especially towards 3GPP (Third
Generation Partnership Project), the key global standard development
organization (SDO) for mobile communications.
5.1 5G PPP Phase 1

Phase 1 performed fundamental research for the 5 th generation of communication
networks. It started early 2015 with the implementation of 19 Projects xxiv, many
of them completing their work around mid-2017, while some continued until mid-
2018. They provided important results on core 5G technologies and managed to
develop solutions that are able to meet nearly all the performance KPIs for 5G.
Phase one projects contributed to 15 core areas of innovation xxv, the “Golden
Nuggets”, hereafter briefly reported.
5G System Design and Evaluation

a) 5G System, Functional, Logical and Physical Architectures
The 5G system aims to provide a flexible platform enabling new business cases
and models integrating vertical industries, such as, automotive, manufacturing,
energy, eHealth, and entertainment. On this basis, network slicing has emerged as
a promising future-proof framework adhering to the technological and business
needs of different industries. In that context, network slicing has been researched
and designed from an end-to-end perspective, spanning over different technology
domains (e.g., core, transport and access networks) and administrative domains
(e.g., different mobile network operators) including management and orchestration
functions. More specifically, slice lifecycle automation was demonstrated with an
architecture, functions and tools that implement cognitive procedures for all
lifecycle phases: preparation phase, instantiation, configuration and activation
phase, run-time phase, and decommissioning phase. Two fundamental
technological enablers include softwarization, e.g., virtualization of network
functions, as well as software-defined, programmable network functions and
infrastructure resources. Other key elements constitute efficient management &
orchestration procedures and protocols, all subject of in depth work. Finally,
scalable, service-centric data analytics algorithms that exploit multi-domain data
sources, complemented with reliable security mechanisms have been addressed for
deploying customized network services with different virtualized NFs (VNF) on a
common infrastructure in a trustworthy manner
At architectural level low latency has received particular attention. It requires data
management (i.e. routed or processed) as close to where it is required i.e. either at
the receiving end or at the source end). A low-latency architecture requires
network intelligence location as close to the edge as possible, such that traffic
which is expected to remain local never needs to travel towards the core of
the network. It in this way minimizes transmission latency. Distribution of
intelligence closer to the edge providing physical, logical and functional
advantages as compared to the more conventional centralized architectural
approach has been defined and characterized including a reference separation of

the Control and User plane.
This also covers a 5G reference security architecture with focus on a logical and
functional architecture. This focus is motivated by general trends such as network
de-perimetrization as well as 5G systems' strong dependency on software defined
networking and virtualization in general. The core of the proposed security
architecture extends and revises the 3GPP security architecture from TS 33.401 to
integrate domain concepts derived from 3GPP TS 23.101 to better support 5G trust
models, going beyond “telecom” and “mobile broadband”. Strata allows
characterizing different functional aspects and security feature groups are used to
describe security objectives.
b) 5G Flexible Radio Access Network (RAN)
This critical part of 5G systems has received particular attention in 5G PPP phase
one, in view of supporting the subsequent heavy work undertaken by 3GPP in this
domain. The work has notably delivered:
- An agile resource management framework, built upon novel 5G aspects, i.e.,
diverse service requirements, slice service level agreements (SLAs), novel
communication modes and dynamic radio topologies, which capitalizes on the
flexible physical layer numerology. Within this framework, the developed
interference management schemes are adaptive to changing radio topologies
supporting movable access nodes. To address requirements of latency-critical
services, dynamic traffic steering is developed for fast data routing on the RAN
side without hard handovers, as opposed to legacy traffic steering schemes.
Besides, RAN moderation schemes ensure the optimum number of active access
nodes to fulfill service requirements while improving network energy efficiency.
- An initial access scheme based on service differentiation. The scheme enables
both lower latency for mission critical services as well as high resource utilization.
It allows disabling the always-on signals and instead use more dedicated signaling
and self-contained transmission.
- mm-wave RAN integration for access, achieved with solutions involving tight
interworking, multi-connectivity capabilities (to guarantee at least one access point
with the desired level of performance) and lower-band integration. In the context
of the overall RAN design, a new User Equipment (UE) state especially suited for
bursty connections has been defined. The new state makes use of the stored RAN
context and can thereby save Radio Resource Control (RRC) signaling which
enables a better UE power consumption and lower latency.
- Low latency performance implementation in the context of demanding
applications such as intelligent transportation systems (ITS) relying on high-speed,
ultra-high reliable and secure digital connectivity. Low-latency as well as
application initialization times has been demonstrated at the physical (PHY) layer,
where a round-trip end-to-end (E2E) latency between UE and central office (CO)
of 6.69 ms has been achieved, using ultra-high speed routing at the node router.
- Instantiation and management of different slices such that each slice may be
potentially orchestrated in a different way, and thus be tailored to the requirements
of a specific service. This requires the coordination of resources between different
slices, which leads to the introduction of a new compound Software-Defined
Mobile Network Control (SDMC) architecture.
- optimized dimensioning of transport requirements for different candidate 5G
RANs, including below 6GHz massive Multiple Input – Multiple Output (MIMO)
and mmWave RANs, and a variety of potential functional splits including L1
processing at the RRH, lower MAC and upper MAC splits.
- design and demonstration of advanced antenna systems for massive MIMO
featuring 96 antenna elements and integrated L1 processing to reduce fronthaul
data rate requirements by a factor of six to twelve, depending on the number of
virtual ports employed.
c) 5G Spectrum Requirements and Candidate Bands
5G PPP has supported the 2016 proposal of the Radio Spectrum Policy Groupxxvi
(RSPG) on 5G “pioneer bands” with the following approach: (i) Low bandwidth
spectrum (700 MHz) to enable 5G coverage to all areas, ensuring that everyone
benefits; (ii) Medium bandwidth spectrum (3.4-3.8 GHz) to bring the necessary
capacity for new 5G services mainly in urban areas; (iii) High bandwidth spectrum
(26 GHz) to give ultra-high capacity for innovative new services, enabling new
business models and sectors of the economy to benefit from 5G. In that context a
joint 5G PPP approach has been worked out in support of the justification for
spectrum beyond 6 GHz, clarifying the expected usage of different bands for
different deployments and services. Different spectrum licensing and sharing
options have also been studied taking into account the innovative properties of
higher frequency bands. This latter part has notably been useful in the context of
the second RSPG opinionxxvii outlining the licensing “toolkit” available to national
regulators.
d) 5G Performance Evaluation Framework
A holistic and programme level 5G RAN performance evaluation framework has
been developed for a numerical assessment of 5G KPIs and notably user data rates,
reliability, control and user plane latencies, mMTC density or energy efficiency,
as well as for comparison of technical solutions proposed for 5G. A visualization
tool to illustrate the potential benefits that mm-wave cell deployment can offer has
been developed with an open source software, and made available to all 5G PPP
projects.
e) 5G Integrated Transport Networks, Fronthaul, Backhaul
Converged fronthaul/backhaul (FH/BH) services solutions based on active and
passive optical technologies are needed to optimize resource of access point
clusters with significantly higher capacity than in the case of 4G. Specifically,
passive WDM-PON technology targets 25 Gbps/wavelength, colorless optical
network unit (ONU) deployments and dynamic ONU switch-off for energy saving.
Active optical technology is based on Time Shared Optical Networks (TSON). An
adaptive, flexible and software-defined architecture for future 5G transport
networks integrating multi-technology fronthaul and backhaul segments has been
achieved. The architecture thus aims to enable a flexible and software-defined
reconfiguration of all networking elements through a unified data plane and control
plane interconnecting distributed 5G radio access and core network functions,
hosted on in-network cloud infrastructure. This work has notably delivered: (i) an
innovative architecture design for 5G transport networks targeting the integration
of existing and new fronthaul and backhaul technologies and interfaces and (ii) a
multilayer data plane architecture, including circuit- and a packet- switched paths.
5G Air Interface Innovations
A new radio interface covering new spectrum frontiers has been a key focus of 5G
PPP early research activities. Main developments have been as follows:
a) 5G Multi-Service Waveform
New waveforms adapted for service coexistence below 6 GHz and overcome the
demerits of Cyclic-Prefix Orthogonal frequency-division multiplexing (CP-
OFDM - the 4G waveform) has been a strong focus in view of addressing poor
spectral containment, lack of robustness in highly asynchronous and high mobility
scenarios, as well as inflexibility for the support of diverse numerology. This was
achieved by applying filtering techniques: subcarrier-wise filtered solutions and
subband-wise filtered solutions. Common to all is the amelioration of spectral
localization of the signal power, which improves the performance particularly for
Massive Machine Communications (MMC), Mission Critical Communications
(MCC) and vehicular (V2X) services and ensures an efficient coexistence of these
services with Mobile Broadband (MBB) service. Whilst CP-OFDM was
eventually adopted at 3GPP level for feasibility reasons, this work is particularly
relevant for future evolutions of 5G radio systems and future releases of the 3GPP
standard.
For services above 6GHz, radio-interface concepts and solution have been
addressed. Twelve mm-wave challenges were analyzed and solution proposed
based on: i) OFDM based waveforms, justified by a high spectral efficiency, easy
integration with MIMO, lower complexity, time localization, and reasonable
robustness to RF impairments, ii) enhancement of Low Density Parity Check
(LDPC) codes and Polar codes with regard high throughput and robustness against
hardware computation imprecision, iii) a flexible frame structure for TDD/FDD
operations considering joint access and backhaul, low latency transmissions, as
well as novel reference signal options to handle high mobility, phase noise, and
CSI acquisition for large antenna arrays, iv) novel schemes for multiple access and
initial access, v) multi-antenna solutions for hybrid beam-forming. Key
components of the developed air-interface solution have been evaluated via
simulator(s) as well as hardware-in-the loop trials. A dedicated channel model
based on a comprehensive measurement campaign has been developed. Most of
the achieved results have been reflected in subsequent 3GPP normative work.
b) 5G Flexible Interference Mitigation and Radio Resource Management

(RRM)
Complementary work has addressed indoor and indoor/outdoor scenarios where
capacity demands are the highest, but also where the proposed extended Dynamic
Spectrum Allocation (eDSA) may be the most effective at exploiting co-operation
across technologies and bands. As a result advanced RRM interacting with higher-
level entities, enabling operator spectrum policy management for all types of
regulatory regime have been designed with flexible and adaptive multi-RAT MAC
for dynamic spectrum access and aggregation.
c) Technology Enablers for 5G RAN Platforms (HW & SW)
Enabling technologies are critical for 5G success. In 5G PPP, an architecture
design for the transceiver of medium range base stations that supports three radio
bands together with a design of a multiband high-power amplifier has been
developed. The presented three-band transceiver solution considers radio bands
defined for mobile communication (E-UTRA band 7 and 38 at 2.6 GHz and band
22 and 42 at 3.5 GHz) and one band between 2.7 and 2.9 GHz, Also, significant
work has been done on context-aware, cognitive and dynamic HW/SW partitioning
algorithm for 5G network elements. This algorithm exploits knowledge (e.g.
prediction of a hotspot) derived by network and sensor measurements and decides
upon the HW or SW execution of functions in order to fulfill and maintain the
application goals. The algorithm leads to high flexibility, performance and energy
efficiency. Moreover, full duplex technology provides gains in the user data rate
of up to 50 % and in aggregated data rates (in a multiuser setting) of up to 21 %.
In addition, integrated management of physical and virtual infrastructures, which
enables automated deployment of 5G infrastructures and services running on top
of them, including virtualisation services, cloud computing, Mobile Edge
Computing (MEC), SDN/NFV services and value-added services such as Service
Function Chaining (SFC) have been addressed. Consequently, the creation and
deployment time for infrastructures and their services are greatly reduced.
d) 5G Massive Channel Access
Commercialization and deployment of 5G systems need to support very high
connection densities to make the Internet of Things serviceable. Massive
connectivity is supported by new air interfaces that should optimise the available
radio and infrastructure resources, spanning areas from protocol enhancements and
radio resource management to waveform design. A new waveform design is
proposed for asynchronous small packet transmissions in the uplink. Also, due to
the superior spectral properties of certain waveforms, the need for tight temporal
synchronization of users can be relaxed. This allows compressing or even avoiding
broadcast messages, thus leading to energy and radio resource savings. In addition,
new, "one-stage" access protocols are being developed, in which access
notification and data delivery are performed in a single transaction by means of
one or more consecutive packets or in a single transmission thereby reducing
signaling overhead for short messages.
Network Management and Security Innovations
Applying advanced technologies and approaches at network management level is
key to reduce OPEX of the target 5G infrastructure. The 5G PPP reached
significant results in the following domains:
a) 5G Network Management
Autonomic network management has been progressed in order to improve
network performance whilst reducing operational expenditures (OPEX). 5G
autonomic network management is powered by artificial intelligence and extends
the current 4G Self-Organising Network (SON) concept in the physical layer to
both 5G physical and virtual domains. The work includes 5G network self-
monitoring which collects and analyses performance metrics at multiple levels:
physical infrastructure, virtual infrastructure and traffic flows with multi-tenancy
awareness, thereby enabling timely situation awareness of 5G network
infrastructures and services. A set of key, high-level Health of Network (HoN)
metrics are modelled and introduced, and example HoN metrics include Virtual
Infrastructure Vulnerability, Cyber-Attack Risk, and Video QoE. These
innovative, customisable and extensible HoN metrics greatly facilitate speedy and
more precise identification of common network problems.
At RAN level, a unified control and coordination framework for 5G
heterogeneous RANs has been designed. It takes advantage of SDN for RAN
programmability, in particular for efficient radio resource modelling and
management, and flexible spectrum management. Major progress includes
flexible RAN architecture design, radio resource abstraction, RAN data models
and application program interfaces (APIs), virtualization and coordination
framework, and flexible spectrum management. In that context, resource
utilisation improvement and support of inter-operator spectrum sharing have been
demonstrated which has led to subsequent take up at 3GPP level.
b) 5G Networks Security and Integrity
A 5G Security Vision has been developed as well as a Technical Roadmap on
security enablers for major areas of concerns (namely AAA, Privacy, Trust,
Security Monitoring and Network management and virtualization) as confirmed
by Open Consultation ran publicly on 5G. The researched security enablers come
with open specifications for anyone interest to come up with its own
implementation and are linked to major building blocks of the 5G Security
Architecture defined and they contribute to. 5G Security enablers when software
released (either open source or closed source based on decision left to enabler
owner) also come with documentation (manuals) to integrate/deploy also make
use of them within the 5G Security testbed according to use terms and conditions
that apply. A 5G test-bed has been designed and set-up to satisfy the requirements
of the 5G security enablers against the threats emerging from identified use cases.
Launched in 2016 and based on three interconnected nodes provided by b<>com,
VTT and Nokia.
Virtualisation and Service Deployment Innovations
Virtualisation and Software implementation of network functions is at the heart
of the 5G promise to offer fully adaptable network environments tailored to the
requirement of the specific tenants. In that context, this area was subject of
massive R&I work under 5G PPP phase 1. Main achievements have been as
follows:
a) Network Softwarization and Programmability integrating SDN and NFV
Technologies
At RAN level, an SDN-based control plane unifying high capacity Point-to-
Multipoint line of sight (P2MP LoS) mmWave radios and below 6GHz non line of
sight (NLoS) radios has been developed. It features Openflow extensions for the
mmWave and below 6GHz radios and common SDN controller implementing
traffic engineering applications for the wireless backhaul, including load

balancing, interference aware scheduling, and fast re-route.
At multi domain level, the split of Network Function Virtualization Orchestrator
(NFVO) into Network Service Orchestrator (NSO) and Resource Orchestrator
(RO) has been worked out as a solution for multi-domain interactions. This split of
functionality has been complemented with a Slice as a Service (SlaaS) approach
for multi-domain RO-RO interworking. While the ETSI NFV framework
architecture has assumed so far that the NSO and RO functions are played by a
single entity (the NFVO), the ETSI NFV has subsequently adopted a similar NSO-
RO functional split for single administration. Initial design and deployment of a
large scale test-bed connecting 13 sites, including 4 operators providing the
connectivity backbone, and emulating realistic Internet topologies today has been
achieved.
Additionally, a flexible NFV MANO Service Platform was developed for NFV
that is built on a micro-service architecture and released as open-source software
under the Apache 2.0 license. The Service Platform operates and manage the
lifecycle of network service on top of a virtual infrastructure manager, like
OpenStack. To this end, it deploys the virtual network functions as virtual
machines and steers the traffic by implementing service function chains. In that
respect, complex services can be built through the “chaining” of these Micro-
VNFs, with different virtualization approaches used to support these micro-VNFs.
b) E2E Orchestration in Single and Multi-Domains 5G Virtualized
Networks
From an end to end perspective and in view of providing multi domain
interoperability and resource “stitching”, the multi-domain orchestration process
has been broken down into the main functions relevant to a multi-provider multi-
domain environment: discovery, bilateral negotiation, provisioning and assurance
stages with their corresponding multi-domain reference points in a detailed multi-
domain orchestration architecture. A proof of concept prototype of the multi-
provider, multi-domain orchestration as an integration of the major concepts from
several projects has been developed with advanced transport control.
Similarly, solutions have been proposed to allow infrastructure owners to
dynamically share (control, manage, orchestrate) their resources, virtual and
physical, in an isolated manner (network slices), and among several network

operators to offer different customized services to their end customers. To that end,
an Open Access Manager (OAM) module has been developed which is responsible
for the lifecycle management of network slices including communications with
network, cloud, or device controllers for creation and operation of a slice and
exposing it to upper layers for dynamic service provisioning.
This is complemented with service and domain aware orchestration which is
responsible, among other functions, for placing virtualized functions in the most
appropriate location. Herein, it takes into account the requirements of the
corresponding service that needs to be satisfied, the constraints on the placement
of functions that interact with each other and the features of the underlying
infrastructure. Orchestration takes place end-to-end, spanning the whole mobile
network from the user to the packet data network respectively to the service
provided for the user. Accordingly, end-to-end orchestration typically involves
multiple stakeholders spanning infrastructure provider, mobile service provider
and tenant, allowing different network slices to have their own Management and
Orchestration (MANO) stack implementation.
c) Programmable Industrial Networks
Applicability of 5G technologies to vertical industries is at the heart of the
European 5G strategy. In that context, SDN & NFV ecosystem, based on open,
modular and secure framework have been implemented to showcase a prototype
for intra-domain and inter-domain scenarios in real wind parks as a representative
use case of industrial networks. The wind-park control network is very
representative of a professional application as wind energy has now established
itself as key for sustainable energy generation. With lower capital expenditure and
operational expenditure costs in control network infrastructure, the validation of
the economic viability of the approach has been demonstrated. Further
applicability in other industrial domains are expected to be demonstrated as follow
up.
d) Flexible and Agile Service Deployment
Fast service deployment is needed in 5G considering the very large number of use
cases and related service requirements that are expected. Service Platform
developments have been complemented with a Service Development Kit (SDK)

that aims at fast implementation, testing, and debugging of virtualized network
functions and services. The SDK supports the creation of function and service
descriptors as well as service packages uploaded to the Service Platform and used
to manage the lifecycle of complex network service. It also offers features to test,
profile, and debug network services locally by using a Service Platform emulator
that mimics the behavior of the actual Service Platform locally, e.g. on a
developer's laptop. The knowledge gained by these local tests simplifies the
function and service development, shorten the time-to-market, and at the same
time, increases the quality of the resulting product. The SDK interconnects tightly
with the Service Platform and allows monitoring running services in real time. To
this end, it enables the developer to collect important data and offers tool to analyze
the data in order to debug or improve the service, for example in terms of
performance.
In complement, a framework to automatically map service level Key Performance
Indicators (KPIs) to the platform level parameters in the host compute environment
have been developed. The framework enables the identification of platform
features, which most significantly influence the KPIs for a given workload under
test.
In conclusion, phase one has addressed a multiplicity of 5G enabling
technologies and has demonstrated their applicability in view of reaching the main
5G target KPI’s, notably at radio level. Table 3 below provides a sample of
achieved results in that domain:
Key Performance
Baseline/Target Demonstrated
Indicator
Data rates 20 Gb/s DL - 10 Gb/s UL (target) 21 Gb/s DL - 12 Gb/s UL
User Plane Latency 1 ms (ITU target) 0,76 ms
User Data Rate 100 Mb/s (LTE) dense urban > 300 Mb/s
Wireless backhaul N/A 4Gb/s at 60 GHz
Cell edge throughput 15 Mb/s LTE average 170 Mb/s
Interference
Static TDD 60% gain with Dynamic TDD
cancellation
Aggregation of Demonstrated with flexible
multiple radio access N/A RRM (radio resource
techs management)
Energy savings LTE RAN savings up to 66%
31 % with integrated
Energy efficiency Current class of devices
antenna/amplifiers
Video transmission corrupted by 600 Km/h perfect video
Quality of service Intercarrier interference at very transmission demonstrated
high speeds with new waveform/MAC
Positioning N/A <1m Proof of concept
400 MHz contiguous with
Channel bandwidth 20 MHz contiguous (LTE) flat response at 26 GHz
implemented in trial
Table 3. Sample radio results achieved in phase 1
5.2 5G PPP Phase 2

Phase 2 capitalises on phase one developments to move towards Proof of concepts
and trials in the context of a multiplicity of vertical use cases. It started in June
2017, with 22 new 5G PPP projectsxxviii complemented by 2 projects dealing with
international collaboration with Taiwan. Most Phase 2 projects terminate in 2019,
whilst a few will continue in 2020.
The phase 2 move towards use cases and the 5G user community has been
particularly successful, considering that 62% of participants in phase 2 projects
were newcomer to 5G and not participating in Phase 1 projects. This is a positive
indication of the 5G interest to the vertical community in Europe.
Phase 2 projects cover eight vertical clusters: Automotive, Energy, Health,
Industry, Media and Entertainment, Public Safety, Smart Cities, Transport and
Logistics. The various projects cover different 5G KPI, as a function of the eMBB,
URLLC, or mMTC requirement they have to satisfy for the considered use cases.
The peak of PoC’s and trials targeting verticals occurs with Media and
Entertainment. One reason for this is that eMBB (enhanced Mobile Broadband)
is among the first services to be supported with early 5G deployment.
Cities are as well an important focus, with many trials and pilots directly or
indirectly contributing to the 5G smart city ecosystem. About 50% of the
implemented projects target eMBB use cases whilst mMTC and URLLC use cases
are covered by about 25% of the projects each. Focussing on Smart Cities, trials
will take place in 2019 and 2020. This timing is compatible with the availability
of technologies that support the 5G New Radio, core network functions and the
user equipment, and is in line with the early trial objectives set out in the 5G
Action Plan. Another important aspect in this context is the provision of radio
spectrum for 5G, which may somehow vary across EUxxix. Altogether, phase 2
projects account for 63 PoC’s and trials in 38 cities across 13 EU countries. Figure
2 below illustrates the typology of experiments and their locations in member
States whilst figure 3 illustrates the city locations for PoC’s and trials across the
8 vertical clusters supported by Phase 2 projects
Figure 2: Typology and location of Phase two POC’s and trials

Figure 3: European cities covered by 5G PPP phase 2 PoC’s and trials
5.3 5G PPP Phase 3

Phase 3 has two objectives: i) to support the development and rollout of 5G
innovation and validation platforms across Europe in view of fostering 5G
acceptability in a wide variety of business sectors from the technological and
business point of view and, ii) to support the European R&I ecosystem to prepare
longer term evolution of 5G and “beyond 5G” systems.
The 5G PPP phase 3 has been structured in steps as follows:
- the setup of and end to end 5G platform that may be used by verticals to run
large scale validation of their use cases, using at least 5G Release 15 compatible
technology, hence with characteristics very close to commercially deployed

platforms. Under this step, 3 projects have been selected and started in June 2018;
- the setup of dedicated large scale pilots for Connected and Automated Mobility,
taking into account the strategic role identified for this vertical at the level of the
5G AP, and supporting the Member States commitments to make available cross
border corridors for 5G based CAM. Another set of 3 projects started in November
under this step
- The implementation of large-scale trials and pilots for amultiplicity of verticals,
running on top of the selected end to end 5G validation platforms. This step is to
be implemented early 2019;
- the launch of a longer term 5G “long term evolution” initiative, in view of paving
the way towards systems beyond 5G with a medium to long term approach and
taking into account technologies not yet addressed under the 5G PPP. This step is
to be implemented in 2019.
The last work programme covering the year 2020 will complement these phase 3
initiatives.
i) 5G end-to-end test and validation platforms.
The complexity of 5G and the multiplicity of technologies integrated under one
single platforms call for the availability of end to end 5G platforms to validate the
vertical use cases of interests. The main objective is to actually demonstrate with
user pilots the core 5G capability, namely that one single infrastructure can
instantiate several networks serving different requirements, even if conflicting,
with shared and isolated resources. In that context, the target is to implement a
multiplicity of concurrent vertical use cases on top of one single end-to-end
platform, such that the overall performance can be tested and validated in close to
operational conditions. This call for availability of equipment with characteristics
that are close to equipment actually deployed in operational networks, and the
selected platforms will first implement Release 15 technology and gradually
upgrade with Release 16 technology as the specifications become available.
The 5G Infrastructure PPP Phase 3 platforms projectsxxx (2018-2021) are:
- 5G EVE
- 5G VINNI
- 5GENESIS
They all started in July 2018 provide large-scale end-to-end 5G validation
network infrastructure, covering about 20 EU sites and nodes on a pan-EU basis.
This infrastructure will provide the adequate level of openness to make it possible
for vertical industries to test their innovative 5G business cases using ad-hoc
network resource control in an end-to-end interoperability framework.
The key platforms and cities of the PPP Phase 3 projects are outlined in figure 4
below.
Figure 4: 5G PPP Phase 3 Platforms Projects – Site distribution
The key capabilities and features are summarized in the Table below. It may be
noted that (1) capabilities will be incrementally added until the end of the projects;
(2) interworking refers to interconnection of two or more sites to provide unified
service(s) in the 5G E2E facility and interconnection does not assure
interworking; (3) integration will be developed with Vertical Pilots projects; (4)
exact time line will be communicated after selection of the vertical projects
running on top of the platforms. The notes (2), (3) and (4) apply as reported in
specific corresponding capabilities rows.
Platforms Capabilities 5G-EVE 5G-VINNI 5GENESIS
Rel15-5GNR in Non Turin, Madrid, Oslo, Kongsberg, Athens, Berlin,

Standalone Alone (NSA) Paris, Athens Martlesham, Limassol,
mode January 2020 Malaga, Surrey
Patras, Madrid,
Aveiro
Rel15-5GNR with Rel15- Turin, Madrid, Oslo, Kongsberg, Athens, Malaga,
5GCore in Standalone Paris, Athens Martlesham, Berlin, Surrey
Alone (SA) mode(4) June 2021 Patras, Madrid,
Aveiro.
After Jan2020
Rel16-5GNR and 5GCore Turin, Madrid, Oslo, Kongsberg, Surrey
(NSA or SA)(4) Paris, Athens Martlesham,
June 2021 Madrid, Patras
After Jan2020
Network Slicing as a - Oslo, Kongsberg, Athens, Malaga,
service(3) Martlesham, Surrey
Patras, Madrid,
Aveiro
Customized network slice Turin, Madrid, Oslo, Kongsberg, Athens, Berlin,
(e.g. SFC, security, Paris, Athens Martlesham, Limassol,
enhanced Cloud access) (3) June 2021 Patras, Madrid, Malaga, Surrey
Aveiro
Hosting of 3rd party Turin, Madrid, Oslo, Kongsberg, Athens, Berlin,
VNFs(3) Paris, Athens. Martlesham, Limassol,
January 2020 Patras, Madrid, Malaga, Surrey
Aveiro
Interworking(2) with other Turin, Madrid, Oslo, Kongsberg, Athens, Malaga,
platform facilities(3) Paris, Athens Martlesham, Surrey
June 2021 Patras, Madrid
Integration of additional Turin, Madrid, Oslo, Kongsberg, Athens, Berlin,
gNB to platform facility(3) Paris, Athens Martlesham, Limassol,
January 2020 Patras, Madrid, Malaga, Surrey
Aveiro
After Jan2020
Edge Computing Turin, Madrid, Madrid, Aveiro, Athens, Malaga,
Paris, Athens Oslo, Kongsberg, Surrey
January 2020 Patras,
Martlesham
(TBD).
After Jan2020
Distributed Data fabric - Oslo, Kongsberg, Athens
service for analytics Patras (TBD),
Madrid.
After Jan2020
3.5 GHz 5G Radio Turin, Madrid, Oslo, Kongsberg, Athens, Malaga,

Paris, Athens Martlesham, Surrey
January 2020 Patras, Madrid,
Munich, Aveiro
26 GHz 5G Radio Oslo, Kongsberg, Surrey
Martlesham
Millimeter wave for Patras Surrey
Backhaul
End UserTesting Turin, Madrid, Oslo, Kongsberg, Athens, Berlin,
Paris, Athens Martlesham, Limassol,
January 2020 Patras, Madrid Malaga, Surrey
Automatic testing Turin, Madrid, Oslo, Kongsberg, Athens, Malaga,
framework Paris, Athens Martlesham, Surrey
Patras, Madrid
Table 3: 5G PPP Phase 3 Platforms Projects – Capabilities/Features/
When implementing vertical use cases large-scale pilots on top of these platforms,
a key objective will be to satisfy the network resource control by multiple tenants
with the adequate level of resource visibility/control required by these tenants.
Another important aspect will be to demonstrate composability of resources
across multiple domains and also the right level of service/data isolation between
the different use cases ran in different slices. These pilots are expected to start
their implementation in the second half of 2019.
i) Pilots over cross border corridors for Connected and Automated
Mobility
Three projects startedxxxi under this initiative:
- 5G CARMEN;
- 5G CROCO
- 5G MOBIX
These projects only cover a subset of the currently existing ten cross border
corridors made available by Member States. 5G CARMEN covers the Brenner
Corridor over IT-AT-DE, 5G CROCO covers the Metz-Merzig-Luxembourg
corridor linking FR-DE-LU; 5G MOBIX covers the Porto-Vigo, Evora-
Merida corridor linking PT-ES and the Greece-Turkey corridor with and 8km
segment across the border. The main target is to leverage vehicle connectivity to
create complete ecosystem around cars, serving not only safety related
applications but a rich blend of application including maintenance, insurance,
infotainment, driver’s assistance and autonomous driving. The projects take a
holistic perspective covering V2x scenarios, namely vehicle to vehicle (V2V),
vehicle to infrastructure (V2I), vehicle to network (V2N), vehicle to pedestrian
(V2P). This latter mode is by far a priority for safety services considering that the
highest number of road fatalities concerns collisions between cars and pedestrians.
In that particular case, the possibility to establish ultra-low latency connectivity
with user smart phones is of the essence.
More generally, these projects are addressing a multiplicity of issues such as:
- Use of hybrid new radio to validate 5G radio in CAM context
- Multi domain service orchestration for cross domain seamless service provision
- Distributed multi layer embedded cloud for low latency and multi service
support;
- Novel type of services paving the way towards Release 16 capabilities such as
remote driving, see through, high density platooning, share my view scenarios as
typical use cases defined by 3GPP.
The projects are also expected to assess the economic feasibility of the considered
scenarios and will evaluate the interworking and co-existence/complementarity
scenarios with safety dedicated technologies such as ITS-G5. In view of
addressing more cross border corridors and complementary innovation, the
Commission proposes complementary CAM actions for the last work programme
of Horizon 2020, with target implementations in 2020. Finally, the European
Commission has proposed CAM follow up actions for operational deployment of
5G technology along main transport paths under the CEF2 xxxii proposals of the
next Multi Annual Financial Framework (MFF).
6 Spectrum and Standards: Key Enablers

As identified in the 5G Action Plan, the early availability of spectrum is critical
to enable 5G deployment in Europe. The identification of 5G spectrum above 24
GHz (i.e. in the millimetre wave range) is notably framed by the ITU preparatory
process, which has identified a number of candidate bands between 24.25 and 86
GHz to be studied until the next World Radiocommunication Conference in 2019

(WRC-19).
Whilst the 28 GHz band was not part of the bands to study by the last WRC-15,
several nations (US, South Korea, Japan) have already decided to assign this band
for early 5G introduction, in view of driving the international spectrum policy
agenda. The US FCC declared its intentions to license the 28 GHz band for 5G
already in summer 2016 under the “Spectrum Frontiers” initiative.
Based on these fast international developments, Europe addressed the 5G
spectrum issue through the Radio Spectrum Policy Group (RSPG). The RSPG
reached conclusions in its Opinion on 5G Spectrum, adopted in November 2016
and supported by the European industry, that there was a need to select a set of
pioneer bands for early 5G trials and pilots, in order to ensure appropriate early
commercial deployment in 2020. The 26 GHz band is one of these pioneer bands,
together with the 700 MHz and 3.6 GHz bands. Having a band in the millimetre
wave range is essential to cater for the very high data speeds targeted by 5G,
thanks to the considerably larger bandwidth it can offer compared to the lower
spectrum bands. Its compatibility for the use by 5G is under study by the CEPT
in view of further spectrum harmonisation initiative of the Commission planned
for 2019. These positive developments do all support the ambitious deployment
objectives of the 5G AP. Amongst the other key ITU members, China is also
promoting the 26 GHz band in the international context, whilst the 3,5 Ghz band
enjoys a very good level of support globally.
In Europe, recent spectrum developments have shown that member States are now
moving forward with assignments of 5G spectrum. Several MS have already
launched auctions, whilst a significant number of auctions are being planned for
2019 or being prepared through national consultations. The Commission through
the European 5G observatoryxxxiii launched in September 2018 consistently
monitors the spectrum status in Europe. As indicated in the context of the launch
of 5G PPP end to end trial platforms, it may be seen that 3,5 GHz and 26 Ghz are
already available in some MS for test and pilot experiments. Early availability of
these bands is indeed key to stimulate the development of a device ecosystem and
to allow ambitious trials with vertical users to be set up. 5G PPP R&I initiatives
will hence actively leverage spectrum availability in leading Member States.
Similarly, global standards are key to support the ambitious use cases targeted by
5G. For the first time in the history of mobile communications, the prospects of
one single 5G global standard is real, considering that all key stakeholders and
nations have joined forces in 3GPP to define a global standard.
3GPP initiated the development of the 5G standard in September 2015, with an
inception workshop in Phoenix, USA. The first version of the standard was
released in June 2018, under the Release 15, with a specification that covers the
Non Standalone (NSA) and Standalone (SA) options. NSA implies that the 5G-
access node relies on a 4G core or on another 4G eNodeB to establish the
connection. This is the deployment option selected for almost all the early
deployment scenarios as it allows operators to capitalise on earlier 4G
investments. The SA option relies on a complete 5G gNode B and a 5G core. Few
operators, notably China Mobile, intend to proceed with SA deployment at this
stage.
The fast development of this first release was pushed by the aggressive
deployment agenda of operators from the US and Asia. Consequently, the first
release has focused on the eMBB use case and on the “New Radio” (NR)
specification, in view of the high rate high capacity options contemplated by these
operators. The standard cover notably:
- The NR functionalities for enhanced mobile broadband (eMBB) and ultra-
reliable low-latency-communication (URLLC) as defined in [TR38.913].
URLLC focus has primarily been on low latency. NR specification consider
frequency ranges up to 52.6 GHz. The NR functionalities have been designed
to be forward compatible and allow for smooth introduction of additional
technology components and support for new use cases. Backward
compatibility of the NR to LTE is not required.
- Layer 1 and Layer 2 user plane specifications include a common part to all
supported architecture options. However and for some aspects, Layer 1 and
Layer 2 user plane specifications are different between single connectivity
and Dual Connectivity options.
- Several connectivity options and scenarios as defined in TR38.801 and
corresponding to either NSA or SA options are considered for this normative
work. These scenarios envisage several configurations to link (or not) to an
existing 4G infrastructure, hence providing maximum flexibility to operators

to implement 5G as a function of their existing 4G deployment status.
The next stage of development of the standard will take place under Release 16
that started in June 2018 for the normative phase. This is illustrated in figure 5
below:
Figure 5: 5G standardization time frame at 3GPP level (source, 3GPP)
Release 16 has two aspects on its agenda:
i) A 5G expansion, the main extensions being:
• Vehicle to X (V2X), moving NR into V2X in complementary

mode
• Industrial IoT
• Ultra-Reliable Low-Latency Communication (URLLC)
enhancements, notably the reliability part which is critical for
mission critical applications
• Unlicensed spectrum operation
• NR-Broadcasting, though only at study stage
• Non terrestrial networks (NTN), namely for the satellite community
ii) At the same time important work is planned to improve 5G

efficiency through:
• Interference Mitigation
• 5G Self-Organising Networks (SON) & Big Data
• 5G Multiple-input multiple-output (MIMO) enhancements
• 5G Location and positioning enhancements (also key for industrial
applications)
• 5G Power Consumption improvements
• Dual Connectivity enhancements
• Device capabilities exchange
Release 16 is hence important to make 5G fully compatible with vertical use cases.
In particular, the work under the “5G expansion” work item relates to important
use cases, notably automotive, Industry 4.0 and factories, but also healthcare and
energy. 5G PPP phase 2 and phase 3 projects will both contribute to and benefit
from these important standardisation developments.
7 Conclusion
The European Commission has clearly identified 5G as a key infrastructure to
fulfil the wider policy objectives aiming at a modernised digital industry and
economy. Bold support has been provided to industry through a structured and
targeted research programme responding to policy initiatives and aiming at
accelerating the availability of 5G in Europe. Moving towards user pilots is now
of paramount importance, and the framework conditions to make this happen are
rapidly developing (availability of technology, frequency bands, standards,
regulations). It is now up to the industry to seize the opportunities and to develop
ambitious business plans to make Europe a lead market of the 5G era.
Acknowledgements
The authors would like to thank the colleagues from the 5G Industry Association,
the 5G PPP technical board and the 5G PPP projects for their huge efforts to make
5G PPP project a coherent set of complementary initiatives delivering a consistent

set of important results. Their efforts in structuring and analysing the overall
contribution and impacts of the projects at programme level have been essential
to provide the overview reported in this paper.
References
i5G Pathfinder project launched under the 7th Framework Programme of the Union:
http://europa.eu/rapid/press-release_IP-13-159_en.htm
ii
5G Manifesto for timely deployment of 5G in Europe: http://telecoms.com/wp-
content/blogs.dir/1/files/2016/07/5GManifestofortimelydeploymentof5GinEurope.pdf
iiiiii
Connectivity for a European Gigabit Society package, 14 September 2016,
https://ec.europa.eu/digital-single-market/en/connectivity-european-gigabit-society
iv
ITU Recommendation M. 2083
v
5GAmericas presentation at 4th 5G Global Event, Seoul, November 2017.
vi https://5g-ppp.eu/wp-content/uploads/2015/02/5G-Vision-Brochure-v1.pdf
vii
Communication from the Commission to the European Parliament, the Council, the European
Economic and Social Committee and the Committee of the Regions: Digitising European Industry
Reaping the full benefits of a Digital Single Market (COM(2016) 180 final):
http://europa.eu/rapid/press-release_IP-16-1407_en.htm
viii
ABI research: https://www.abiresearch.com/press/expanding-beyond-mobility-management-
enterprise-mo/
ixIdentification and quantification of key socio-economic data to support strategic planning for
the introduction of 5G in Europe SMART 2014/0008, study, studying automotive, health,
transport and energy sectors. https://ec.europa.eu/digital-single-market/en/news/identification-
and-quantification-key-socio-economic-data-strategic-planning-5g-introduction
x
Study on the 5G business potential :
http://www.5gamericas.org/files/7114/9971/4226/Ericsson_The_5G_Business_Potenti
al.pdf
xiEricsson mobility report 2016: https://www.ericsson.com/assets/local/mobility-
report/documents/2016/ericsson-mobility-report-november-2016.pdf
xii
Communication from the Commission to the European Parliament, the Council, the European
Economic and Social Committee and the Committee of the Regions: Connectivity for a
Competitive Digital Single Market - Towards a European Gigabit Society - COM(2016)587:
https://ec.europa.eu/digital-single-market/en/news/communication-connectivity-competitive-
digital-single-market-towards-european-gigabit-society
xiii
IMT Vision - "Framework and overall objectives of the future development of IMT for 2020 and
beyond" Rec M2083, https://www.itu.int/rec/R-REC-M.2083
xiv
White Paper "5G Innovations for new business opportunities": https://5g-ppp.eu/wp-
content/uploads/2014/02/5GPPP-brochure-final-web.pdf
xv https://5g-ppp.eu/wp-content/uploads/2014/02/5G-PPP-5G-Architecture-WP-July-2016.pdf
xvi
https://5g-ppp.eu/wp-content/uploads/2014/02/Advanced-5G-Network-
Infrastructure-PPP-in-H2020_Final_November-2013.pdf
xvii
https://5g-ppp.eu/our-members/
xviii
https://www.networld2020.eu/
xix
https://5G PPP.eu/wp-content/uploads/2018/01/5G PPP-5G-Architecture-White-
Paper-Jan-2018-v2.0.pdf
xx
https://5G PPP.eu/wp-content/uploads/2014/02/5G PPP_White-Paper_Phase-1-
Security-Landscape_June-2017.pdf
xxi
https://5G PPP.eu/wp-content/uploads/2017/03/5GPPP-brochure-final-web-
MWC.pdf
xxii
https://5G PPP.eu/wp-
content/uploads/2017/03/NetworkManagement_WhitePaper_1.pdf
xxiii
https://5G PPP.eu/wp-content/uploads/2014/02/5G
PPP_SoftNets_WG_whitepaper_v20.pdf
xxiv
https://5g-ppp.eu/5g-ppp-phase-1-projects/
xxv
https://5g-ppp.eu/phase-1-key-achievements/#
xxvi
http://rspg-spectrum.eu/2016/11/rspg-opinion-on-5g-adopted/
xxvii
http://rspg-spectrum.eu/rspg-opinions-main-deliverables/
xxviii
xxix
http://5gobservatory.eu/5g-spectrum/national-5g-spectrum-assignment/
xxx
xxxi
ibid
xxxii
https://ec.europa.eu/commission/publications/connecting-europe-facility-
digital-europe-and-space-programmes_en
xxxiii
http://5gobservatory.eu/
5G as an opportunity for SMEs
Nicola Ciulli and Jacques Magen1
Abstract If properly prepared, the potential disruption in business models both in

the telecommunications sector and in the various vertical markets where 5G
technologies could be used in the future could prove to be a great opportunity for
Small and Medium-Sized Enterprises.
1 The European 5G Public-Private Partnership

The 5G Infrastructure Public Private Partnership (5G PPP) is a joint initiative
between the European Commission and European ICT industry (ICT
manufacturers, telecommunications operators, service providers, SMEs and
researcher Institutions). Started in 2014, the 5G PPP delivers solutions,
architectures, technologies and standards for the ubiquitous next generation
communication infrastructures of the coming decade. Its main challenge is to
secure Europe’s leadership in the particular areas where Europe is strong or where
there is potential for creating new markets such as smart cities, e-health, intelligent
transport, education or entertainment & media. The 5G PPP will reinforce the
European industry to successfully compete on global markets and open new
innovation opportunities.
Nicola Ciulli
Nextworks, Member of the Board of the 5G Infrastructure Association
Jacques Magen
InterInnov, Member of the Board of the 5G Infrastructure Association
54 Nicola Ciulli, Jacques Magen
The 5G PPP has already been playing an essential role in the development of
5G technologies (Phase 1) and ensuring the feasibility of deploying those
technologies in various vertical sectors (Phase 2). With the advent of Phase 3,
starting in 2018 i.e. now, the European Commission and the European industry,
along with academic organizations, are looking on the one hand at deploying 5G
test infrastructures, and on the other hand at experimenting 5G-based solutions in
various vertical sectors, with the objective to speed up the commercialization of
5G technologies and 5G-based solutions, thus ensuring European global
leadership.
2 SMEs in the 5G landscape

Small and Medium-sized Enterprises (SMEs) employ 93 million people in
Europe, account for 67% of the total employment in the EU-28 non-financial
business sector, and generate 57% of value added, again in the EU-28 non-
financial business sector. From 2013 to 2016, EU-28 SME employment grew by
5.2%, almost 50% faster than overall employment in the EU-28 economy over the
same period. Even in the EU-28 manufacturing sector, in which large enterprises
are generally dominant, SMEs still account for 58% of total employment and 42%
of total value added. In the business services sector, where most of the ICT work
is performed nowadays, SMEs account for 82% of total employment and 76% of
value added [1].
SMEs have an important role to play in developing, piloting and deploying
5G technologies, both to help with disruptive technologies and to address the
needs of various vertical sectors. European SMEs have a great added value in
providing innovative concepts and solutions that are having an important impact
on the 5G value chain. They have the agility and flexibility required in a fast
evolving technical and market landscape. This has been well understood within
the 5G PPP, where a specific objective was set from the start, for SMEs to reach
a 20% share in the programme. The 5G Infrastructure Association (5G IA), which
represents the private side of the 5G PPP, also recognized the importance of SME
involvement: since 2017, the Board of the 5G Infrastructure Association includes
two SME representatives out of 11 members, instead of one previously.
5G as an opportunity for SMEs 55
The involvement of SMEs in the 5G PPP has been progressing along the
Phases, almost reaching 20% in Phase 2. Now, with Phase 3 and the actual 5G
experimentations starting, SME involvement should again increase, as there are
many SMEs already present in vertical markets such as media, transportation,
energy, smart cities, public safety, just to mention a few.
2 5G: a potential great opportunity for SMEs

The deployment of 5G and its introduction in many vertical sectors could prove
very disruptive not only for the stakeholders currently involved in the telecoms
sector, but also for the players of the vertical sectors. Their business models might
need to be reshaped to consider a completely new paradigm. This might prove to
be an opportunity for European SMEs, who are by nature more flexible and may
be less reluctant to change than larger incumbent players. This may be hopefully
leading to large growth and global leadership for those SMEs who will be the first
ones to take this path i.e. using 5G technology to develop new and innovative
solutions and applications in the vertical markets they already know well, or to
take advantage of their expertise in 5G to develop disruptive solutions to reach
new markets.
This evolution needs to be planned now, to be best prepared when it is time
for 5G-based solutions to be commercialized and deployed. Demonstrating the
feasibility for SMEs to develop and deploy 5G innovative solutions in vertical
markets should allow other SMEs to follow a similar path, eventually gaining
European market share on the global scale. This will strengthen the chances that
European SMEs achieve eventually sustainable growth and become key players
in the new business environment created by the 5G paradigm.
Facilitating the future involvement of many European SMEs in the

development of 5G related solutions and applications in various vertical
sectors should be one of the key objectives of the 5G PPP and more
generally of the public and private sectors.
56 Nicola Ciulli, Jacques Magen
If this is prepared properly, then there is a chance that European SMEs may
prove to be major beneficiaries in the potential economic disruption brought by
5G, and the consequent deployment of innovative 5G-based solutions and
applications in many vertical sectors, thus ensuring a European leadership in many
vertical sectors.
References
1. European Union. Nov. 2017. Annual Report on European SMEs 2016/2017. Focus on self-
employment. SME Performance Review 2016/2017
(https://www.eubusiness.com/topics/sme/sme-report-16-17).
Part II
What is 5G as of today
Functional Architecture
Andrea Detti
1 Introduction
Before 5G, the design of cellular network architecture was centered on the end
users. The end users were the first-class citizens of the network and, generation
by generation, cellular network designers strove to offer them an always-increasing
capacity, thereby transforming the architecture into a full-IP one, as in the case of
4G.
The network architecture was made up of nodes, such as eNB, SGW, P-GW,
or MME for 4G. Each node was usually composed of proprietary hardware and
software, coupled in a single device.
5G is a sort of breakthrough in this thirty-year design pattern. Besides expected
improvements in speed and time (20 Gb/s and 1 ms delay over the air), 5G architecture
introduced a new category of first-class citizens: the verticals. 5G provides commu-
nication services not only for end users, but also for different vertical markets, such
as automotive, energy, city management, government, healthcare, manufacturing,
and intelligent transport systems.
Such heterogeneity creates demand for a level of service agility typical of a
software environment, rather than an "ossified" hardware one. For this reason, 5G
architecture has been designed to allow (and foster) a possible softwarization of net-
work functions. Consequently, software defined networking (SDN), network function
virtualization (NFV) and cloud computing are fundamental technologies for making
full use of the power of a 5G network.
Andrea Detti
CNIT - Electronic Eng. Dept., University of Rome Tor Vergata, e-mail: andrea.detti@uniroma2.
it
59
60 Andrea Detti
1.1 5G network architecture and services
A 5G network is composed of a 5G access network (AN) and a 5G core network

(5GC) [1] (fig. 1). The access network itself is made up of a new-generation radio
access network (NG-RAN) [3], which uses the 5G new radio interface (NR) [4],
and/or a non-3GPP AN (e.g. WiFi, xDSL, etc.) connecting to a 5G core network.
The different network entities are connected by an underlying TCP/IP transport
network, which supports diff-serv QoS.
Fig. 1 5G Network Achitecture
Like previous generations, a 5G network connects user equipment (UE) to external

data networks. The 5G connectivity service is named PDU Session. From a transport
point of view, a PDU session is made by a sequence of NG tunnels in 5GC, and
of one or more radio bearers on the radio interface. This set of "pipes" eventually
connects the UE to its control functions and to the external data network for user
traffic exchange (fig. 2). A major task of the mobile network is to establish and release
the tunnels and the bearers dynamically, so as to follow user movements and states
(idle, connected, etc.).
A PDU session is very similar to an EPS bearer in LTE, except for the QoS
model and the supported user data units. Indeed, a PDU session can transport not
only user plane IP packets, but also Ethernet or unstructured frames, thus allowing
layer-2 communication among groups of UE. The 5G QoS model is based on the new
concept of QoS flow [1], where a flow is the finest granularity of QoS differentiation.
Different QoS flows may belong to a single PDU Session1.
Fig. 3 shows the splits between the 5G functions executed in the NG-RAN and
in the 5G core. In broad terms, the NG-RAN takes care of establishing, maintaining
and releasing the parts of the PDU sessions that cross the radio interface. It copes
1 We note that in the case of LTE, the finest QoS granularity is the EPS bearer. Different QoS
services require different EPS bearers
Functional Architecture 61
Fig. 2 PDU Sessions and QoS Flows: User Plane (source [4])
with physical impairments (e.g., fading, interference, power reduction); inter-gNB

handovers; and session multiplexing (scheduling). The 5GC functions manage the
remaining parts of the PDU sessions and take care of all the other processes not
related to radio access (e.g., mobility management, security, IP address allocation,
etc.).
Fig. 3 Functional Split Between NG-RAN and 5GC (source [4])
1.1.1 New Generation Radio Access Network (NG-RAN)
As shown in fig. 4, the NG-RAN consists of a set of 5G base stations, called gNBs,
which are connected to the 5GC through a set of logical interfaces. As in LTE, gNBs
62 Andrea Detti
can be interconnected through the Xn interface to improve mobility (e.g., handover)

and management functions (e.g., inter-cell interference coordination).
The functionality of a gNB is sometimes distributed. In that case, the resulting
architecture is formed by a central unit (gNB-CU) that controls one or more dis-
tributed units (gNB-DU) through the F1 interface. A distributed unit is connected
to a remote radio head (RRH), i.e., the actual radio transceiver. The central unit is
again split in two parts, one for control plane functions (gNB-CU-CP) and one for
user plane functions (gNB-CU-UP), following the control and user plane separation
(CUPS) / SDN approach already introduced in the latest LTE releases.
Fig. 4 Overall NG-RAN Architecture
Fig. 5 NG-RAN Protocol Stack (original source [4])
Fig. 5 shows the stack of the protocols crossing the radio interface and their
placement on the aforementioned gNB units. The stack is almost the same as the
LTE one, except for the service data adaptation protocol (SDAP) of the user plane.
The main functionalities of the different layers are as follows:
• The physical layer (PHY) contains the digital and analogue signal processing
functions that the mobile and base station use to send and receive information. It
is based on OFDMA, with adaptive carrier spacing (15,30,60,120,240 kHz) and

an adaptive modulation/coding scheme (e.g., from π/2 BPSK to 256 QAM)[2].
• The medium access control (MAC) protocol provides low-level control of the
physical layer, primarily by scheduling data transmissions between the mobile
and the gNB.
• The radio link control (RLC) protocol ensures reliable delivery of data streams
that need to arrive intact (HARQ). It also handles segmentation.
• The packet data convergence protocol (PDCP) carries out higher-level transport
functions related to header compression and security.
• The service data adaptation protocol (SDAP) maps the interaction between the
packet of a QoS flow and a data radio bearer (due to the new QoS framework) by
marking the user data packets properly.
• The radio resource control (RRC) is the signaling protocol used in "access stra-
tum" procedures involving the mobile and the gNB. It includes connection estab-
lishment and release functions; the broadcast of system information; radio bearer
establishment, reconfiguration and release; RRC connection mobility procedures;
paging; and power control.
• The non-access stratum (NAS) protocol is the signaling protocol used between the
UE and the 5GC for PDU session management, security, mobility management,
etc. The 5GC entity that takes care of controlling the UE is the access and mobility
management function (AMF), which is similar to the LTE MME.
1.1.2 5G Core Network (5GC)
To some extent, the NG-RAN architecture, as well as its protocol stack, is similar to
the LTE one. However, the architecture of the 5G core network is unique in many
ways.
The decomposition of the functions executed by the network nodes of the previous
generations led to a 5G architecture completely defined in terms of network functions
(NF) that are exposed as services. Accordingly, as we can see in fig. 6, every block
name ends with the letter "F": function.
As occurs in the NG-RAN, we have a control and user plane separation. In the
user plane, we have one or more user plane functions (UPFs), which mainly carry
out packet forwarding between the different NG-U tunnels (fig. 2) that form the PDU
session. All other network functions belong to the control plane.
Another radical change from the previous generations is the interface modeling,
which has moved from "bit-oriented point-to-point" to "web-oriented service-based."
Indeed, 5G core is said to have a service-based architecture; wherever applicable,
procedures (i.e., the sets of interactions between network functions) are defined as
services, so that it is possible to reuse them.
There is a standardized point-to-point interface (real or logical) between any pair
of interacting 2G, 3G and 4G network entities, and this interface uses a specific
bit-oriented protocol. In the 5GC, the interactions among control plane entities use
64 Andrea Detti
Fig. 6 5G System Architecture, Non-roaming (original source [1])
service-based interfaces, supported by web-oriented tools such as HTTP/2, REST

and JSON.
Where are the differences? Whereas a point-to-point interface connects two well-
defined entities (e.g., LTE S6a is strictly between MME and HSS), a service-based
interface exposed by an entity is actually an API that any other entity could use: it is
a one-for-all product.
Service-based modeling strongly improves the agility of the network in evolving
or adapting itself to unforeseen needs. In the point-to-point interface model, if the
system designers wish to add a new network entity and connect it to a set of N
old network entities, they need to standardize N new interfaces and create related
protocols. This complexity often leads to an ossification of the network. With the
service-based interface model, the designers just have to standardize the API of
the new network entity. Similarly, let us assume that there is a chain of network
functions, namely NFa-NFb. Let us now assume that the designers wish to introduce
another network function, NFc, in the middle of the chain, creating the sequence
NFa-NFc-NFb. With the point-to-point model, they would need to standardize two
new interfaces: NFa-NFc and NFc-NFb. By using the service-based model, they
only need to standardize the API of NFc, and that is only if NFc is a new entity. If
NFc is already standardized, they just need to reconfigure the sequence of functions.
The upper part of fig. 6 shows the set of network functions that form the 5G
control plane. All of them expose service-based interfaces. For this reason, they are
depicted as being connected by a network bus rather than by point-to-point links.
The interface name is equal to the function name, with an "N" used as prefix. In this
arrangement, one NF queries a network repository function (NRF) to discover and
enable communication with other NFs. The insertion of a new network function,
including a third-party one, is merely the insertion of a record in the NRF database.
Under suitable security controls (i.e., authentication and authorization), a subset of
service-based interfaces can be easily exposed to external users, such as third-party

application providers, to enable them to optimize their services.
In the lower part of fig. 6, we have the set of network entities belonging to the
user plane. As we can see, we still have point to point-interfaces there, identified by
an "N" plus a number.
Let us conclude this section by describing the main roles of 5G NFs and their
relation to 4G.
• The user plane function (UPF) handles the NG-U tunnel forwarding and the
related data path services, such as anchoring for handover, QoS, and traffic policy
enforcement. There can be multiple UPFs associated with a UE; these UPFs can
be located in a single slice or in multiple ones. The UPF contains parts of the 4G
SGW and PGW functionalities.
• The session management function (SMF) is the control part of a PDU session. That
is, it configures NG tunnels, allocates IP addresses with DHCP, and configures
traffic steering (e.g., towards a third party or an edge cloud). There can be multiple
SMFs associated with a UE, though only one per slice. The SMF contains parts
of the 4G MME and PGW functionalities.
• The access and mobility management function (AMF) handles all the 5GC signal-
ing coming from and going to the UE. Unlike the SMF, it is a single function that
is present in multiple slices. It supports user access to the network and manages
mobility by interacting with the UE and with other NFs (e.g., SMF, AUSF, etc.).
The AMF contains part of the 4G MME functionality.
• The authentication server function (AUSF) supports authentication for 3GPP and
non-3GPP access. It contains part of the 4G HSS functionality.
• The unified data management (UDM) function can be considered a repository for
UE-related information, such as credentials, identifiers, AMF details, and SMF
assignments for the current session. The underlying idea of the UDM is to create,
wherever possible, a central database for UE configuration information, so that
the NFs can be designed as stateless services, improving architectural agility. The
UDM contains part of the 4G HSS functionality.
• The policy control function (PCF) is a unified entity providing policy rules (QoS,
filtering, charging, etc.) to other control plane functions, such as SMF. The PCF
contains part of the 4G PCRF functionality.
• The network slice selection function (NSSF) selects the set of network slice
instances serving the UE, along with the best AMF for that purpose. It is not
present in 4G.
• The network exposure function (NEF) exposes the capabilities of networks and
network/UE events for third-party, application function, edge computing, and
other purposes. It is not present in 4G.
• The network repository function (NRF) discovers network function instances.
When it receives an NF discovery request from a NF instance, it provides the
discovered NF instances. It is not present in 4G.
• The application function (AF) resembles an application server that can interact
with the other control-plane NFs. AFs can exist for different application services,
and can be owned by the network operator or by trusted third parties. For instance,
66 Andrea Detti
the AF of an over-the-top application provider can influence routing, steering its

traffic towards its external edge servers.
1.1.3 Network Slicing in the 5G Architecture
The 5G core architecture is made of network functions. This structure enables its
immediate deployment with software and cloud tools. Indeed, 5GC is a cloud-native
architecture.
Network functions can be implemented as pieces of software embedded in light
virtual machines (e.g., Docker or Unikernel) and executed using a cloud infrastructure
whose servers are spread all over the 5G network and are interconnected by an
agile SDN. This allows for the easy reconfiguration of virtual network connectivity
among "virtual" NFs. By using such a cloud-based deployment, there is a complete
decoupling of the NFs from both the execution hardware and the interconnecting
network infrastructure.
Cloud-based deployment of the 5G network also makes it possible for a tenant
to create an isolated ICT environment, formed by specific instances of control and
user plane NFs, supported by a dedicated 5GC virtual network and customized radio
bearers. Such an isolated environment is actually a 5G slice, i.e., a network-as-a-
service offered to the different verticals.
As shown in fig. 8, a network operator can deploy multiple network slices with
different features, or with the same features but for different groups of UE. For
example, a slice for its customers can be equal to another slice for the customers of
a virtual operator.
Each slice has a unique identifier, which includes the slice/service type (SST),
referring to the expected behavior of the slice in terms of features and services. Cur-
rently, there are three standardized SST values (fig. 7). These are used to support the
roaming use cases for the most commonly used slice/service types more efficiently.
Fig. 7 Standardised Slice/Service Types (source [1])
As reported in fig. 9, NFs in different slices can be used in different configurations,

or be placed further from or closer to the UE, depending on the vertical application
using the slice. For instance, an eMBB slice could use a high-capacity radio bearer
and have two UPFs, one in the edge and one in the cloud, to better support user
mobility (two anchor points). A vehicular slice could have a radio bearer with low
delay and medium capacity and many control functions moved to the edge to further
reduce latency. An IoT slice could have a low-bit-rate radio bearer, a single UPF
assuming low mobility, and most control NFs in the core, provided that latency is
not important. Although they are not included in the figure, some slices can share
the same instances of NFs, and there are some NFs, such as NSSF, that are common
to all slices.
Fig. 8 5G Slicing Example (source [5])
Fig. 9 5G NFs Deployment in Slices

68 Andrea Detti
References
1. 3GPP TS 23.501: "System Architecture for the 5G System; Stage 2", version 15.3.0 Release
15
2. 3GPP TS 38.221: " NR; Physical channels and modulation ", version 15.3.0 Release 15
3. 3GPP TS 38.401: "NG-RAN; Architecture description",version 15.3.0 Release 15
4. 3GPP TS 38.300: "NR; Overall description; Stage-2", version 15.3.1 Release 15
5. Frank Mademann, "System architecture milestone of 5G Phase 1 is achieved",
3GPP news, available at http://www.3gpp.org/NEWS-EVENTS/3GPP-NEWS/1930-SYS_
ARCHITECTURE
3GPP NR: the standard for 5G cellular networks
Abstract The next generation of cellular networks (5G) will be characterized by

ultra-high data rates, ultra-low latency and support for a massive number of connec-
tions. The 3rd Generation Partnership Project (3GPP) has recently standardized the
specifications for NR, the new Radio Access Network (RAN) designed to match 5G
requirements thanks to the use of the millimeter wave (mmWave) spectrum, massive
Multiple Input Multiple Output (MIMO), a flexible design of the air interface and
novel deployment paradigms. This chapter reviews the main novelties that NR intro-
duces with respect to 4G cellular networks, with a focus on how they can be used to
provide unprecedented performance in 5G deployments.
1 Introduction
According to the International Telecommunication Union (ITU), the 5th generation

(5G) of cellular networks will need to address the traffic demands and new use cases
of the digital society beyond 2020 [17]. In particular, 5G networks should support:
(i) a user experienced rate of at least 100 Mbit/s, with a peak data rate in ideal
conditions of 20 Gbit/s, and three times higher spectral efficiency with respect to
4G; (ii) ultra-low latency, i.e., 1 ms round-trip over the air; (iii) support for mobility,
with communications at up to 500 km/h; (iv) an area capacity of 10 Mbit/s/m2 with
up to 106 connections per km2 ; and (v) a 100x increase in energy efficiency with
respect to 4G networks.
Long Term Evolution (LTE) is the set of specifications that the 3rd Generation
Partnership Project (3GPP) has introduced in 2009 and evolved since then to satisfy
the current 4G requirements. The evolutions of LTE will match some of the next
generation requirements in specific deployment scenarios [12], but they will not be

Department of Information Engineering, University of Padova, Via G. Gradenigo, 6/B, 35131
Padova, e-mail: {polesemi,giordani,zorzi}@dei.unipd.it
69
70 Michele Polese, Marco Giordani, Michele Zorzi
able to effectively address all the 5G use cases. For example, LTE operates with a
maximum of 20 MHz per carrier, thus limiting the achievable data rate, and has a
rigid frame structure that makes it difficult to reduce the round-trip latency below
1 ms. Moreover, LTE has not been designed to account for energy efficiency (e.g.,
pilot signals are always-on) and to support a massive number of connections (even
though this is targeted by the recent Narrow Band IoT (NB-IoT) evolution).
In order to overcome the limitation of LTE networks, the 3GPP has recently
defined a new Radio Access Technology (RAT), i.e., 3GPP NR1, that introduces
novel designs and technologies that will comply with the 5G requirements. NR
exploits a new spectrum, i.e., the millimeter wave (mmWave) band, and features
the support for new techniques such as massive Multiple Input Multiple Output
(MIMO), flexibility in terms of frame structure, to target different use cases, and
multiple deployment options for the Radio Access Network (RAN). Moreover, a
new core network design (i.e., 5G Core (5GC)) has been introduced to offer network
slicing and virtualization, and different deployment options and inter-networking
with LTE have been specified.
NR has been standardized by 3GPP with a first set of specifications2 (Release 15)
in December 2017 and a complete one published in June 2018. Release 16 for NR
is expected to be completed in December 2019, and will be composed of a set of
specifications that match the ITU 5G requirements previously described [3].
This chapter is organized as follows. In Sec. 2 we will describe the main charac-
teristics of the NR air interface design, describing why it is flexible and lean and how
low latency is achieved. Then, in Sec. 3 we will introduce mmWave communications
and describe the procedures integrated in NR for the support of the mmWave bands.
In Sec. 4, we will present massive MIMO techniques, focusing on both below- and
above-6 GHz NR use cases. In Sec. 5 we provide insights on the new RAN and core
deployment options and in Sec. 6 we conclude the chapter.
2 A flexible and lean design
The main characteristic of the NR physical layer is its flexibility: the standard, indeed,
provides a general technology framework designed to address the different and, in
some cases, conflicting 5G requirements [17] and to be forward compatible, so that
it can accommodate future applications and use cases.
Both LTE and NR use the Orthogonal Frequency Division Multiplexing (OFDM)
modulation, which divides the available time resources in frames of 10 ms with
subframes of 1 ms, and frequency resources in subcarriers with spacing ∆ f . More-
over, subframes are further divided in slots and symbols, where the combination
1 While NR was originally meant as the acronym for “New Radio” [2], according to the latest 3GPP
specifications [8] it has lost its original meaning and it now refers to the 5G Radio Access Network.
2 The specifications for NR are in the Technical Specification (TS) of 3GPP 38 series, together
with Technical Reports (TRs) that contain related studies. Other relevant RAN specifications can
be found in the 36 (LTE) and 37 (LTE-NR inter-networking) series.
3GPP NR: the standard for 5G cellular networks 71
Frame – 10 ms
Physical Resource Block

Subframe – 1 ms
12 subcarriers
Bandwidth ! − max 400 MHz per carrier
Examples of slot configurations
with different subcarrier spacings
Slot – 0.25 ms
Subcarrier
spacing
60 kHz
Slot – 0.125 ms
Subcarrier
spacing
120 kHz
Symbol – 8.9 μs
Fig. 1 Frame structure configuration for the physical layer of 3GPP NR
of a single OFDM symbol and a single subcarrier constitutes the smallest physical
resource in NR. While with LTE the symbol duration and the subcarrier spacing
are fixed, with NR it is possible to configure different OFDM numerologies3 on
a subframe basis, i.e., every subframe is self-contained and can be characterized
by a different numerology [6]. This makes it possible to address different 5G use
cases with a single RAT: for example, a shorter OFDM symbol duration, combined
with a higher subcarrier spacing, can be used for high-data-rate and low-latency
traffic, while lower subcarrier spacing can be used for low-frequency narrowband
communications for machine-generated traffic [22]. Fig. 1 illustrates an example of
NR frame structure with two different possible subcarrier spacings.
Another main NR novelty with respect to LTE is the support for ultra-low latency
communications [13], to target the sub-1 ms round-trip latency requirement of 5G.
First of all, the usage of larger subcarrier spacings and shorter symbols has the po-
tential to reduce the transmission time with respect to the basic LTE frame structure.
Moreover, control information related to modulation and resource allocation can
be added at the beginning of data packets, allowing the devices to start decoding
as soon as they start receiving data [6]. This also translates into tighter processing
constraints in 5G NR devices, which must be able to process a received packet in a
few hundreds of microseconds (the actual constraints depend on the subcarrier spac-
ing, as discussed in [12]). Another consequence is that the devices will be able to
transmit the Hybrid Automatic Repeat reQuest (HARQ) acknowledgement after just
one slot, making it possible to reduce the round-trip latency below 1 ms.4 Moreover,
latency-sensitive data does not need to wait for a new slot to be transmitted, but
3 The term numerology refers to a set of parameters for the OFDM waveform, such as subcarrier
spacing and symbol duration [30].
4 In LTE (Release 14), the round-trip latency was fixed to 3 ms [28].
the base station may decide to transmit it as soon as possible using mini-slots, i.e.,
groups of at least 2 OFDM symbols that can be allocated to a data transmission and
do not need to be aligned with the beginning of a standard slot [9].
Finally, in order to increase the flexibility and the energy efficiency of the RAN,
NR limits the number of always-on reference signals, thereby configuring them to
match the deployment scenario and increase the energy efficiency [21]. Moreover,
the self-contained subframe and the minimization of always-on signals make the NR
design forward-compatible, i.e., they enable the evolution of the NR RAT to support
unforeseen use cases with novel technologies and solutions without compromising
the support for legacy devices [9].
3 A new spectrum frontier: millimeter waves in 3GPP NR
5G cellular systems introduce unprecedented requirements in terms of data rate,

latency, link resilience, and end-to-end reliability, which go beyond what exist-
ing mobile technologies can support. In this perspective, the mmWave spectrum –
roughly above 10 GHz5 – has rapidly emerged as an enabler of the 5G performance
demands in micro and picocellular networks [25]. These frequencies, combined with
high-order modulation, offer much more bandwidth than 4G/LTE systems operating
in the congested bands below 6 GHz, and initial capacity estimates have suggested
that networks operating at mmWaves can offer orders of magnitude higher bit-rates
than legacy cellular networks. Moreover, mmWave systems operate through highly
directional communications which tend to isolate the users and deliver reduced in-
terference. Additionally, inherent security and privacy is also improved because of
blockage and of the short-range transmissions which are typically established.
Motivated by the above introduction, NR will boost the 5G performance by sup-
porting, for the first time, frequencies up to 52.6 GHz in Release 15, including
therefore mmWave bands [1]. Nevertheless, communication at mmWaves introduces
new challenges for the whole protocol stack, which may have a significant impact on
the overall end-to-end system performance. First, signals propagating in the mmWave
spectrum suffer from severe path loss and susceptibility to shadowing, thereby pre-
venting long-range omnidirectional transmissions. Second, mmWave links are highly
sensitive to blockage and have ever more stringent requirements on electronic com-
ponents, size, and power consumption. Third, directionality requires precise beam
alignment at the transmitter and the receiver and implies increased control overhead.
In order to overcome these limitations, the NR specifications include new Physical
(PHY) and Medium Access Control (MAC) layer operations to support directional
communications, which are collectively referred to as beam management according
to the 3GPP terminology [15, 8]. In particular, NR networks must provide a mech-
anism by which User Equipments (UEs) and Next Generation Node Base Stations
5 Although strictly speaking mmWave bands include frequencies between 30 and 300 GHz, industry
has loosely defined it to include any frequency above 10 GHz.
frequency
Synchronization Signal (SS) block
Measurements derived from CSI 239

Each SS block is mapped to a signals can be used to maintain the 182
subcarriers
certain angular direction alignment for connected UEs
PBCH
PBCH
PSS
SSS
(directional measurements are
made based on the quality of the 56
received signal).
CSI-RS
CSI-RS
CSI-RS
…
CSI-RS
TCSI TCSI
0
0 1 2 3
Δ#
OFDM symbols
Δ"
CSR-RSs are associated with
time and frequency offsets
… according to the corresponding …
SS burst
SS burst periodicity (TSS) SS burst

time
Fig. 2 Beam management structure in NR systems. SS blocks and CSI-RSs are used for beam
measurements in idle and connected modes.
(gNBs)6 regularly identify the optimal directional beams to interconnect at any given
time.
The following four beam management operations are defined:
• Beam sweeping, i.e., exhaustively scanning a spatial area with a set of beams
transmitted and received according to pre-specified intervals and directions.
• Beam measurement, i.e., measuring the quality, e.g., in terms of received power
(RSRP) or Signal to Interference plus Noise Ratio (SINR), of the received beam-
formed signals [7].
• Beam determination, i.e., selecting the optimal beam (or set of beams) for estab-
lishing directional (and fully beamformed) communications.
• Beam reporting, i.e., the procedure through which the nodes feed back to the
RAN information on the quality of the beamformed signals and on the decision
made in the previous phase.
For idle users, beam management is fundamental to design a directional ini-
tial access strategy, which allows the mobile terminals to establish a physical link
connection when first accessing the network [16]. In this case, beam management
operations rely on a directional version of the 3GPP LTE synchronization signals
called Synchronization Signal (SS) block, i.e., a group of 4 OFDM symbols in time
and 240 subcarriers in frequency with the Primary Synchronization Signal (PSS),
the Secondary Synchronization Signal (SSS) and the Physical Broadcast Channel
(PBCH). Each SS block is mapped to a certain angular direction so that directional
measurements can be made based on the quality of the received signal. To reduce
the impact of SS transmissions, up to 64 SS blocks can also be grouped into the first
5 ms of an SS bursts [6], as illustrated in Fig. 2.
For users in connected mode, as the dynamics of the mmWave channel imply
that the directional path to any cell can deteriorate rapidly, beam management is
required to maintain precise alignment of the transmitter and receiver beams as the
6 gNB is the 3GPP terminology for a base station.

cm
1.5
λ/2
Fig. 3 Illustration of an UPA MIMO array. At the gNB side, the array (which has dimension of
roughly 1.5 cm ×1.5 cm) is comprised of 4 × 4 elements, at the UE side the array has 2 × 2 elements.
The antenna element radiation pattern is modeled as a patch antenna element with horizontal and
vertical spacing equal to λ/2.
UEs move, an operation that is defined as tracking [23]. In this case, besides SS
blocks, Channel State Information - Reference Signals (CSI-RSs) can also be used
for beam measurement operations.
The beam management performance for both idle and connected UEs is a function
of several parameters, including beamwidth, frame structure, SS burst and CSI-RS
periodicity, and gNB density: the trade-off involves network reactiveness, system
overhead and measurement accuracy. In general, better performance can be guaran-
teed considering Non-Standalone (NSA) deployments [5] (which is part of the 3GPP
Rel. 15 standard specifications), in which NR gNBs use LTE as a radio overlay for
control plane management [15].
4 Massive MIMO: a core component of NR systems
While the combination of extreme cell densification, increased system bandwidth,

and more flexible spectrum usage (e.g., by resource sharing) represents a feasible
and sustainable solution to meet 5G performance requirements, MIMO techniques
have also emerged in modern wireless networks to improve reliability and spectral
efficiency. The main concept is to use multiple transmit and receive antennas to
exploit multipath propagation. Among the possible antenna array designs, the most
suitable approach is the use of Uniform Planar Arrays (UPAs) where the antenna
elements are evenly spaced on a two-dimensional plane and a 3D beam can be
synthesized by adapting both azimuth and elevation planes, as illustrated in Fig. 3.
Depending on the channel proprieties, at the PHY layer MIMO systems can be
configured for [29, 26]:
• Spatial diversity, i.e., sufficiently separated antennas are used to transmit redun-
dant versions of the same message over multiple paths. The quasi-independent
fading characteristics of the channel are thereby exploited to make links more
robust and decrease the outage probability.
• Beamforming, i.e., multiple antenna elements are adaptively phased to form a
concentrated beam pattern towards a specific direction. Beamforming provides
significant array gains, thereby guaranteeing increased SNR (since propagation
path loss is mitigated) and reduced co-channel interference (resulting from the
spatial selectivity of the directional antenna).
• Spatial multiplexing, i.e., an outgoing signal is split into multiple independent
streams which are transmitted simultaneously and in parallel on the same channel
through different antennas. Throughput gains can be achieved, provided that
Channel State Information (CSI) is available.
Moreover, multi-user MIMO (MU-MIMO) can be enabled through Spatial Di-
vision Multiple Access (SDMA), in which the multipath proprieties of the channel
are used to multiplex users in the spatial dimension while operating in the same
time-frequency resource.
Typical MIMO installations use relatively few (i.e., less than 10) antennas, and the
corresponding improvement in spectral efficiency has been relatively modest [27].
When combined with mmWave propagation, instead, the full potential of the MIMO
paradigm can be truly unleashed. In fact, the physical size of antennas at mmWave
frequencies is so small that it becomes practical to build large antenna arrays (e.g.,
with 100 or more elements), thereby scaling up the network performance by possi-
bly orders of magnitude compared to state of the art MIMO implementations. The
concept of using a number of antennas in network nodes which is much higher than
the number of users is usually referred to as massive MIMO [18]. The promise of
these benefits has elevated massive MIMO to a central position in NR, with a fore-
seen role of providing high-capacity and almost ubiquitous coverage in ultra-dense
deployments [8]. For mmWave transmissions, massive MIMO is mainly used for
beamforming while, at sub-6 GHz, it provides channel hardening, i.e., the combined
usage of a massive number of antennas decreases the channel variability by averaging
the small-scale fading [11].
However, massive MIMO comes with its own set of challenges, mainly related to:
• hardware impairments: massive MIMO systems exploit channel reciprocity to
estimate the channel responses on the uplink and use such information for both
uplink and downlink transmissions. Since the transceiver hardware is generally
not reciprocal, calibration is needed to exploit the channel reciprocity in practice.
• energy-consumption vs. flexibility trade-off: while it is desirable to design digital
beamforming architectures (which enable the transceiver to generate beams in
multiple directions at the same time), they may suffer from increased energy
consumption with respect to an analog strategy (which, in turn, has little flexibility
since the transceiver can only beamform in one direction at a time).
• CSI acquisition: dynamic environments impose a finite coherence interval during
which CSI must be acquired and utilized. As a consequence, there is a finite num-
ber of orthogonal pilot sequences that can be assigned to the network terminals.
Reuse of such pilots may result in pilot contamination and coherent interference,
which cause performance degradation.
For NR, support for massive MIMO is introduced by using high-resolution CSI
feedback and uplink Sounding Reference Signals (SRSs) targeting the utilization
of channel reciprocity (e.g., twelve orthogonal demodulation reference signals are
specified for multi-user MIMO transmission operations). Additionally, NR focuses
on the support of distributed MIMO, through which the NR devices can receive
multiple independent Physical Downlink Shared Channels (PDSCHs) per slot to
enable simultaneous transmissions from multiple points to the same receiver.
5 Towards a disaggregated and virtualized network
As mentioned in Sec. 2, NR has been designed with flexibility in mind, in order to

address the different 5G use cases. This has an impact also on the possible cellular
network deployment architectures [4, 10], which follow two recent emerging technol-
ogy trends: disaggregation and Cloud RAN (C-RAN) [20], and virtualization [10].
Fig. 4 represents the main novelties in terms of architectures and deployment options
for NR and 5GC.
The LTE RAN and the associated core networks (Evolved Packet Core (EPC))
are characterized by the deployment of standalone pieces of equipment and servers,
e.g., the evolved Node Bases (eNBs), and the core elements such as the Packet
Gateways (PGWs) and Mobility Management Entities (MMEs). With NR, instead,
the gNB can be split into separate physical units, i.e., the Distributed Unit (DU),
which contains the lower layers of the protocol stack and is deployed in the field,
and the Centralized Unit (CU) incorporating complete stack functionalities, which
can be co-located with the DU or hosted in a data center facility, according to the
C-RAN paradigm. As discussed in [19], this allows network operators to deploy the
5G RAN according to the use cases they want to serve, e.g., an ultra-dense small cell
deployment with low utilization but high peak rate can rely on the C-RAN CU/DU
split to maximize the multiplexing and enable a centralized control of the RAN,
while a rural low-density deployment for the support of Internet of Things (IoT)
NSA deployment SA deployment

4G EPC 5G Core
CU
DU DU
CU PGW/SGW AMF SMF PCF
AMF SMF PCF
CU AMF SMF PCF
DU MME UPF
UPF
CU UPF
HSS
DU
Network slicing and NFV
gNB
Radio Access Network Core Network
Fig. 4 Representation of different deployment and operation modes for the 5G NR RAN and the
5GC.
applications can feature complete gNB nodes. Moreover, in order to smooth the
transition between the different network generations and reuse the widely deployed
LTE and EPC infrastructure, the NR specifications foresee a NSA deployment, in
which NR gNBs are connected to the EPC, possibly with a Dual Connectivity (DC)
setup aided by LTE [24]. The other option is a standalone (SA) deployment, in which
both the RAN and the core network respect the 5G specifications.
Finally, the 5G core network has been redesigned with respect to the 4G core
following a service-based approach [10]: the 5G core is composed of multiple
network functions, that provide mobility, authentication and routing support, that
can be dynamically instantiated in data centers according to the load and traffic
demands of the network. For example, while in LTE/EPC networks the control plane
for the mobility of the user was handled by a single server (e.g., the MME), with
the 5GC multiple network functions concur to offer the same set of services, but
can be deployed in different data center locations and quickly turned off and on
to decrease resource utilization. Moreover, the 5GC supports network slicing [14],
i.e., the resources of the network can be split to serve different portions of traffic,
that have different Quality of Service (QoS) requirements (e.g., IoT and mobile
broadband traffic). The service-based 5GC architecture is an important enabler of
network slicing in 5G, given that network functions can be provisioned dynamically
to serve new network slices without the need to use separate servers, as would happen
with the EPC.
6 Conclusions
In this chapter we described the main novelties that the 3GPP has specified for 5G
NR, focusing on how they are used to match the 5G performance requirements. We
showed that 3GPP NR is a flexible technology framework, which can be tuned to
enable a wide range of 5G scenarios: it exploits a novel portion of the spectrum, to
increase the throughput, a frame structure that can provide ultra-low latency, massive
MIMO and several deployment architectures.
References
1. 3GPP, “Study on New Radio (NR) Access Technology - Physical Layer Aspects - Release
14,” TR 38.802, 2017.
2. ——, “TR 38.913, Study on Scenarios and Requirements for Next Generation Access Tech-
nologies, V14.1.0,” 2017.
3. ——, “WF on Work plan of Self Evaluation SI,” RP-172101, 3GPP TSG RAN77, Sapporo,
Japan, September 2017.
4. ——, “NG-RAN; Architecture description - Release 15,” TS 38.401, V15.3.0, 2018.
5. ——, “NR - Multi-connectivity - Overall description (Stage 2),” TS 37.340, 2018.
6. ——, “NR - Physical channels and modulation - Release 15,” TS 38.211, V15.0.0, 2018.
7. ——, “NR - Physical layer measurements - Release 15,” TS 38.215, 2018.
8. ——, “NR and NG-RAN Overall Description - Release 15,” TS 38.300, 2018.
9. ——, “Study on New Radio (NR) access technology,” TR 38.912, V15.0.0, 2018.
10. ——, “System Architecture for the 5G System; Stage 2 - Release 15,” TS 23.501, V15.3.0,
2018.
11. E. Björnson, E. G. Larsson, and T. L. Marzetta, “Massive MIMO: ten myths and one critical
question,” IEEE Communications Magazine, vol. 54, no. 2, pp. 114–123, February 2016.
12. E. Dahlman, S. Parkvall, and J. Skold, 5G NR: The next generation wireless access technology.
Academic Press, 2018.
13. R. Ford, M. Zhang, M. Mezzavilla, S. Dutta, S. Rangan, and M. Zorzi, “Achieving Ultra-
Low Latency in 5G Millimeter Wave Cellular Networks,” IEEE Communications Magazine,
vol. 55, no. 3, pp. 196–203, March 2017.
14. X. Foukas, G. Patounas, A. Elmokashfi, and M. K. Marina, “Network Slicing in 5G: Survey
and Challenges,” IEEE Communications Magazine, vol. 55, no. 5, pp. 94–100, May 2017.
15. M. Giordani, M. Polese, A. Roy, D. Castor, and M. Zorzi, “A Tutorial on Beam Management
for 3GPP NR at mmWave Frequencies,” IEEE Communications Surveys & Tutorials, 2018.
16. M. Giordani, M. Mezzavilla, and M. Zorzi, “Initial access in 5G mmWave cellular networks,”
IEEE Communications Magazine, vol. 54, no. 11, pp. 40–47, November 2016.
17. ITU-R, “IMT Vision - Framework and overall objectives of the future development of IMT
for 2020 and beyond,” Recommendation ITU-R M.2083, September 2015.
18. E. G. Larsson, O. Edfors, F. Tufvesson, and T. L. Marzetta, “Massive MIMO for next generation
wireless systems,” IEEE Communications Magazine, vol. 52, no. 2, pp. 186–195, February
2014.
19. A. Maeder, A. Ali, A. Bedekar, A. F. Cattoni, D. Chandramouli, S. Chandrashekar, L. Du,
M. Hesse, C. Sartori, and S. Turtinen, “A scalable and flexible radio access network architecture
for fifth generation mobile networks,” IEEE Communications Magazine, vol. 54, no. 11, pp.
16–23, November 2016.
20. N. Makris, C. Zarafetas, P. Basaras, T. Korakis, N. Nikaein, and L. Tassiulas, “Cloud-Based
Convergence of Heterogeneous RANs in 5G Disaggregated Architectures,” in IEEE Interna-
tional Conference on Communications (ICC), May 2018, pp. 1–6.
21. S. Parkvall, E. Dahlman, A. Furuskar, and M. Frenne, “NR: The New 5G Radio Access
Technology,” IEEE Communications Standards Magazine, vol. 1, no. 4, pp. 24–30, Dec 2017.
22. N. Patriciello, S. Lagen, L. Giupponi, and B. Bojovic, “5G New Radio Numerologies and
their Impact on the End-To-End Latency,” in IEEE 23rd International Workshop on Computer
Aided Modeling and Design of Communication Links and Networks (CAMAD), Sept 2018,
pp. 1–6.
23. M. Polese, M. Giordani, M. Mezzavilla, S. Rangan, and M. Zorzi, “Improved Handover
Through Dual Connectivity in 5G mmWave Mobile Networks,” IEEE Journal on Selected
Areas in Communications, vol. 35, no. 9, pp. 2069–2084, Sept 2017.
24. M. Polese, M. Mezzavilla, and M. Zorzi, “Performance Comparison of Dual Connectivity and
Hard Handover for LTE-5G Tight Integration,” in Proceedings of the 9th EAI International
Conference on Simulation Tools and Techniques. ICST (Institute for Computer Sciences,
Social-Informatics and Telecommunications Engineering), 2016, pp. 118–123.
25. T. S. Rappaport, S. Sun, R. Mayzus, H. Zhao, Y. Azar, K. Wang, G. N. Wong, J. K. Schulz,
M. Samimi, and F. Gutierrez, “Millimeter wave mobile communications for 5G cellular: It
will work!” IEEE Access, vol. 1, pp. 335–349, 2013.
26. S. Sun, T. S. Rappaport, R. W. Heath, A. Nix, and S. Rangan, “MIMO for millimeter-wave
wireless communications: beamforming, spatial multiplexing, or both?” IEEE Communica-
tions Magazine, vol. 52, no. 12, pp. 110–121, December 2014.
27. A. L. Swindlehurst, E. Ayanoglu, P. Heydari, and F. Capolino, “Millimeter-wave massive
MIMO: the next wireless revolution?” IEEE Communications Magazine, vol. 52, no. 9, pp.
56–62, September 2014.
28. K. Takeda, L. H. Wang, and S. Nagata, “Latency Reduction toward 5G,” IEEE Wireless
Communications, vol. 24, no. 3, pp. 2–4, June 2017.
29. D. Tse and P. Viswanath, Fundamentals of wireless communication. Cambridge University
Press, 2005.
30. A. A. Zaidi, R. Baldemair, H. Tullberg, H. Bjorkegren, L. Sundstrom, J. Medbo, C. Kilinc, and
I. D. Silva, “Waveform and Numerology to Support 5G Services and Requirements,” IEEE
Communications Magazine, vol. 54, no. 11, pp. 90–98, November 2016.
Use of millimeter wave carrier frequencies in 5G
Stefano Tomasin
Abstract Electromagnetic signals with carrier lengths between 1 mm and 1 cm are

commonly denoted as mm-waves, and they are going to be used in the fifth gener-
ation (5G) of mobile communications. This chapter will outline the main features
of mm-waves transmissions. We first discuss the drivers and opportunities of using
mm-waves in 5G mobile systems, underlining technical aspects and peculiar phys-
ical phenomena. Then we introduce signal processing techniques that are essential
for mm-waves, indicating also relevant issues for 5G systems, namely: channel es-
timation, hybrid beamforming and initial access. We conclude the chapter with a
description of most promising applications and some notes on regulatory aspects of
5G mm-wave systems.
1 Drivers and Opportunities for 5G Systems
An ambitious performance is expected from the fifth generation (5G) of cellular

communication systems in order to accommodate a variety of applications (e.g.,
remote control, monitoring, intelligent transport systems, and tactile interaction),
with user experienced data rates up to 1 Gbps (500 Mbps) in downlink (uplink) and
latency as low as 0.5 ms. These targets can not be met by using only the spectrum
available for 4G systems1, but require the use of new frequencies. When the interna-
tional telecommunication union (ITU) started as early as in 2012 the international
mobile telecommunication 2020 (IMT-2020) standard2, it also pushed for a new
spectrum allocation to cellular systems worldwide. The world radiocommunication
Stefano Tomasin
Department of Information Engineering, University of Padova, via G. Gradenigo 6/B, 35131 Padova,
Italy, e-mail: tomasin@dei.unipd.it
1 In 4G systems the center frequencies are from 1 to 3 GHz, and the total bandwidth is up to 10 GHz.
2 IMT-2020 will be defined by third generation partnership project (3GPP) as a release, finally
commercialized as 5G system to the end user.
79
80 Stefano Tomasin
conference (WRC) in 20153 has identified various spectrum portions between 24 to

86 GHz for mobile communications. Tens of GHz are expected to be made available
in the mm-wave band, according to specific spectrum assignments made by each
country.
The use of these frequencies for mobile communications opens various technolog-
ical opportunities and challenges. Among opportunities, mm-waves are particularly
favorable to the 5G scenario of small-cells envisioned by ITU, as their strong at-
tenuation naturally reduces interference. Moreover, the maturity of complementary
metal-oxide semiconductor (CMOS) transistor production has allowed a cheaper im-
plementation of the circuitry needed to handle millimeter wavelength signals. On the
other hand, mm-waves also pose technical channels to overcome strong attenuations,
provide a good channel estimate, and discover new users entering cells, as described
in the following.
Before proceeding, we recall that, strictly speaking, the portion of spectrum
carrying signals with wavelength between 1 mm and 1 cm (thus being more entitled
to be named mm-waves) is denoted by ITU as the extreme high frequency (EHF)
band and spans frequencies from 30 to 300 GHz. However, even frequencies below
30 GHz, down to 6 GHz (about 5 cm wavelength) are typically denoted as mm-waves,
at least in a 5G context.
2 Physical Characteristics
The most important characteristic of the mm-wave band for radio communication
is its significant path loss attenuation due exclusively to the distance between the
transmitter and the receiver. Indeed, the specific attenuation in free space due to the
atmosphere raises from 5·10−3 dB/km at 2 GHz to about 2·10−1 dB/km at 24 GHz to
dramatically increase to 20 dB/km at 60 GHz, due the peak of absorption of oxygen:
these numbers translate into an attenuation of 10 dB at 200 km, 50 km, and 500 m,
respectively. In order to incorporate other propagation phenomena, various channel
models have been proposed for mm-waves (see [1] for an overview).
Moreover, the mm-waves are subject to the blockage phenomenon, as their prop-
agation is largely prevented by almost any physical object. This phenomenon is
typically captured by either shadowing, modelling the presence of static objects, or
fading, accounting for fast attenuations variations due to moving objects.
Within the 3GPP 5G standard the channel model is described in the technical
report series TR 38.901 [2], which specifies various scenarios, including urban
macro and micro cells, rural macro cells and indoor office.
3 The WRC is held every three or four years to review and revise radio regulations, the international
treaty governing the use of the radio-frequency spectrum and the geostationary-satellite and non-
geostationary-satellite orbits.
Use of millimeter wave carrier frequencies in 5G 81
2.1 Integrated Circuits Advancements
The spectrum of mm-waves has been used for many years by astronomers and
military, thus we may wonder why engineers working on mobile communications
become truly interested in this portion of the spectrum only in the ’10s of the third
millennium. Together with an economic push to obtain new spectrum that would
allow higher data rates, we find also technological reasons.
The signals transmitted by smartphones and other consumer devices are generated
and amplified by integrated circuits typically built with silicon-based technologies,
like CMOS or bipolar CMOS (BiCMOS). Any active device has a maximum fre-
quency of operation, called transit frequency, beyond which it does not provide any
current gain. Practical amplifiers can actually be built only at a fraction of the transit
frequency. As the transistor transit frequency is inversely proportional to its geo-
metrical size, the continuous downscaling of the microelectronic technology favored
systems operating at increasingly higher frequencies. Only in ’10s of the third mil-
lennium the industry has been ready to mass-produce transistors with a minimum
feature down to 28 nm and below, providing a reliable platform for applications
operating above 10 GHz.
Indeed, the advance of this technology has pushed the use of mm-waves in many
fields, including communications (devices for HD transmission from digital set top
boxes, Wi-Fi standards, satellite communications), sensors (road radars for cars,
body scanners), and medical applications.
3 Signal Processing and Protocols
Due to the peculiarities of the mm-wave channel, special signal processing techniques
and protocols must be adopted. Some surveys on mm-waves and their use for 5G
communications have appeared in recent years and provide further insight into these
and other topics [3, 4, 5]. Here we outline three relevant issues: the use of multiple
antennas; the estimation of the channel characteristics; and the discovery of new
users entering a cell.
3.1 Massive MIMO and Beamforming
The use of multiple antennas at both the transmitter and the receiver is an effective
means to overcome the attenuation of the mm-wave channel. In fact, by sending
the same signal over multiple omnidirectional antennas with proper delays and
amplifications we ensure that the signals add up coherently in the specific position
where the receiver is. The overall effect is to create a directional antenna which
sends data much farther away in specific directions than a single omnidirectional
antenna that disperses power in all directions. The reason of using different delays
82 Stefano Tomasin
and amplifications per antennas is related to the fact that the distances between
each transmit antenna and the receiver are different, thus determining arrivals with
different delays and attenuations, which must be compensated before transmission.
This transmission strategy is named transmit beamforming.
Similarly, when at the receiver multiple antennas are available, the signals col-
lected by them can be suitably delayed and combined in order to obtain a stronger
replica of the transmitted signal. Note that while the same data signal is received by
all receive antennas, in general independent zero-mean noise samples are present on
each antenna, that are averaged out by combining. This is named receive beamform-
ing.
A communication system with multiple transmit and receive antennas is a multiple
input multiple output (MIMO) systems. The more antennas are used, the more
focused is the beamforming: a system with a very large (from hundreds above)
number of antennas is a massive MIMO system. However, we can also use multiple
transmit and receive antennas to transmit simultaneously multiple signals, each
carrying different data in what is called multiplexing transmission.
Let us now consider the baseband representation of a MIMO system. In a nar-
rowband transmission between NT transmit and NR receive antennas, sk (nT) is the
baseband signal at time nT on transmit antenna k = 1, . . . , NT , where T is the sam-
pling time and n is the time index. Similarly, at the receiver rm (nT) is the sampled
signal at time nT on antenna m = 1, . . . , NR after demodulation. In the baseband
representation, the effect of delays and attenuations of the channel is represented by
a matrix multiplication of vector4 s(nT) = [s1 (nT), . . . , s NT (nT)]T by the NR × NT
matrix H with complex entries, providing the input-output relation
r(nT) = [r1 (nT), . . . , r NR (nT)]T = H s(nT) + w(nT) , (1)
where w(nT) = [w1 (nT), . . . , w NR (nT)]T is a vector of NR independent zero-mean

complex additive white Gaussian noise (AWGN) variables with variance σw2 . The
beamforming is obtained by digitally transforming the transmitted and received
signal vectors. In particular, let H = U DV H be the singular value decomposition
(SVD) of H, where D is a M × M diagonal matrix, M = rankH ≤ min{NR, NT }, U
and V H are unitary matrices of size NR × M and M × NT , respectively, with
U H U = I NR , V H V = I NT , (2)
and IK is the identity matrix of size K. Then by transmitting s(nT) = V d(nT), with
d(nT) a vector of M data symbols, and multiplying the received vector r(nT) by
U H , from (2) we have
d̃(nT) = U H r(nT) = U H HV d(nT) + U H w(nT) = Dd(nT) + w̃(nT) , (3)
where w̃(nT) is a vector of M independent zero-mean AWGN variables with vari-

ance σw2 . Therefore the MIMO channel has been converted into M parallel AWGN
4 T denotes the transpose operator. H denotes the Hermitian operator.

channels, as (3) can be written as d˜m (nT) = Dm dm (nT) + w̃m (nT), m = 1, . . . , M,

where d˜m (nT), dm (nT) and w̃m (nT) are the m-th entries of vectors d̃(nT), d(nT)
and w̃(nT), respectively, and Dm is the m-th entry of the diagonal of D. By this
beamforming procedure at the same time we simplify the implementation of the
transmitter and the receiver and maximize the data rate of the transmission, i.e., we
achieve the capacity of the MIMO channel. Note that from the definition of M the
number of antennas at either side of the communication system limits the number of
data that can be simultaneously transmitted. In practice, experimental results have
provided reduction of the attenuation by 40 dB with the transmitter and the receiver
equipped with uniform planar array antennas with 64 elements [6].
3.2 Channel Model
Although the number of transmit and receive antennas can be large, the different
paths from the transmitter to the receiver are related to the objects surrounding the
devices, and they do not increase with the number of antennas. Moreover, mm-waves
are subject to strong absorptions upon hitting most objects, thus only few reflections
occurs before the signal is completely absorbed. Taking into account the strong free-
space attenuation, typically the number of paths L over which a signal travels from
a transmitter to a receiver is L ≈ 3.
In this very simple scenario a simple model is available for the channel matrix H,
taking into account the multiple antenna geometry. Consider for example uniform
linear arrays (ULAs) where antennas are uniformly spaced with δ spacing along a
line and indicate with λ the carrier wavelength. The length of path l = 1, . . . , L from
the transmit to the receive antennas is xl . Assuming that xl >> δ, the departure angle
θ l(T) is the angle between the line of the transmit antennas and the line connecting
the transmitter to the reflecting object. Similarly path l is characterized by the arrival
0.5
δ 0.4
δ cos(θ1(R) ) 0.3
0.2
0.1
x1 0
100 100
50 50
0 0
RX beam number TX beam number
(a) Single path received by a (b) Example of channel inverse discrete Fourier
ULA. transform (IDFT) of 3GPP channel.
Fig. 1 Example of mm-Wave received signal.

84 Stefano Tomasin
angle θ l(R) between the line of the receive antennas and the line connecting the
receiver to the reflecting object. Fig. 1.a shows the incidence of the single path on a
ULA with NR = 3 antennas. Then, the channel matrix has entries
L
Õ l l
[H] p,q = αl e2π jηR p e2π jηT q , p = 0, 1, . . . , NR − 1, q = 0, 1, . . . , NT − 1, (4)
l=1
xl
δ δ
where ηRl = λ cos θ l(R) , ηTl = λ cos θ l(T) and αl = 1 −j2π
xl2
e λ . The statistics of each
parameter depend on the considered propagation scenario, and various relevant cases
can be found for example in the 3GPP mm-wave channel model [2].
In conclusion, although the channel matrix H is large, it turns out to be defined
by very few parameters, namely, the angles of arrival and departure and the length of
L paths. These features can be exploited to simplify some procedures connected to
the use of massive MIMO technology, such as channel estimation and initial access.
3.3 Hybrid Beamforming
For the technology of the ’10s of the third millennium, digital to analog converters
(DACs) are quite expensive and power consuming components. When NT grows
large, the NT digital to analog conversions of vector s(nT) represent a problem.
Therefore, it has been suggested to perform beamforming partially in the digital
domain and partially in the analog domain. In fact we remember that multiplying
a baseband signal by a complex number corresponds to attenuating and "delaying"
the corresponding analog narrowband signal. Therefore, the matrix multiplication
V d(nT) can be performed in the analog domain by scaling, delaying and adding
analog signals according to the entries of V . For a transmit beaforming completely
performed in the analog domain we need M DACs to convert the M symbols of vector
d(nT) instead of the NT >> M DACs required to fully digital represent V d(nT).
On the other hand, the analog part should be reconfigurable since the channel
matrix changes for different users and different scenarios. In practice, only a finite
(small) set of delays are practically implementable, thus introducing quantization
effects in the beamforming process. In order to rape the benefits of both analog and
digital approaches, a significant effort has been done to develop hybrid beamforming
structures, where a first digital matrix operation (beamforming) is performed on
d(nT) generating an intermediate vector y(nT) of size M̃ with NT > M̃ > M.
Then y(nT) is converted into an analog signal and further processed by an analog
beamforming structure to generate the NT signals to be transmitted. Fig. 2 shows a
general hybrid beamforming scheme.
Various solutions have been proposed, and for a survey on hybrid beamforming
for 5G system refer to [7]. We only note that from (4) we have that for a large number
of antennas matrix U can be written in part as a Fourier transform matrix that can
be easily implemented both with analog circuits and with digital operations [8].
y1 s1
Analog Beamforming
Digital Beamforming
DAC
d1
dM
y M̃ s NR
DAC
Fig. 2 Example of transmit hybrid beamforming structure.
3.4 Channel Estimation
In order to perform beamforming, matrix H must be known. The procedure used

to obtain H goes under the name of channel estimation. In order to estimate it at
the user terminal, the base station (BS) transmits a sequence of NT pilot symbols
represented by the NT × NT matrix S, known also by the user, which collects NT
samples from each antennas. By exploiting the linear relation R = H S + W , where
we collected NT received columns vectors r into R, and solving the linear system
in H we obtain the least squares (LS) estimate of H, H̃ = RS −1 , still affected by
AWGN noise with variance σw2 independent at each channel matrix entry, if S is
unitary.
The LS method works for any MIMO system, and does not exploit the peculiarities
of the mm-wave channel, where matrix H can be described by a reduced number of
parameters and has the structure given in (4). In order to exploit these peculiarities
we consider the virtual channel, or angular domain representation. This is obtained
by taking the two-dimensional M1 × M2 Fourier transform of matrix H as
NR Õ NT L
1 Õ 2π f i j
− 11 − 22
2π f i j Õ
[G] f1 , f2 = [H]i1 ,i2 e M1 e M2 = αl [W (Ωl )] f1 , f2 , (5)
M1 M2 i =0 i =0 l=1
1 2
where fi = 0, 1, . . . , Mi − 1, i = 1, 2, W (Ωl ) is a matrix containing the two-

dimensional (2D) sampled periodic sinc function5 centered at frequencies Ωl =
(Ωl1, Ωl2 ) = (M1 ηRl , M2 ηTl ) . Fig. 1.b shows an example of mm-wave channel in the
5 The two-dimensional (2D) sampled periodic sinc function is defined as
NR ( f1 + Ω1l ) j π( f1 +Ωl ) NR −1 NT ( f2 + Ω2l ) j π( f2 +Ωl ) NT −1

" ! # " ! #
NR NT
[W (Ωl )] f1 , f2 = I NR e 1 M1 × I NT e 2 M2 ,
M1 M1 M2 M2
sin(π x)
where In (x) = n sin( π nx )
is the 1D-periodic sinc function.
86 Stefano Tomasin
angular domain: we can clearly distinguish a few peaks, corresponding to the small
number of paths L.
The reason to move into this transformed domain is that the 2D-sinc functions that
compose G are concentrated among their peaks, and the effect is more remarkable
for a large number of antennas. In other words, from a full matrix H we obtain
a sparse matrix G. Note also that by the Fourier operation, the noise statistics are
not altered, thus all points of the LS estimate of G are still affected by independent
Gaussian noise.
In order to refine the LS channel estimate we find the peaks in the estimate of
G and then reduce (or set to zero) all other values of the matrix that contain only
noise. Lastly, by an inverse Fourier transform we obtain a new estimate of H with
much reduced noise. Various solutions have been proposed for this purposes, and
[5] provides a survey. For example, solutions based on the iterative detection and
cancellation of the paths from the virtual channel estimation have been proposed in
[8], also under the name of orthogonal matching pursuit (OMP). The sparsity of the
channel can also be exploited by compressed sensing approaches, for example using
fast iterative shrinkage-thresholding algorithm (FISTA), least absolute shrinkage and
selection operator (LASSO), basis pursuit denoise (BPDN) and accelerated gradient
descent with adaptive restart (AGDAR) [9] algorithms.
3.5 Initial Access
When a user enters a new cell, the user and the BS must find each other in space by
turning their beamformers in the proper direction, in what denoted as initial access
problem.
The basic solution provides that both BS and user sweep the space, the BS trans-
mitting a pilot signal and the user detecting its direction of arrival. This corresponds
approximately to estimating the strongest path of the mm-wave channel. Then the
user transmits a pilot signal towards the BS that sweeps the space until it detects its
direction of arrival.
This procedure can be refined in various ways (see [10] for a survey), in order
to reduce the time by which a new user is discovered. For example, by exploiting
positioning information coming from external sources (e.g., from a global navi-
gation satellite system receiver) and exchanging this information over an existing
communication channel at a lower frequency, the beamforming direction to be used
at both terminals can be inferred. However, various phenomena can make the beam-
forming direction different from the line of sight: for example, obstructions may be
present, while a good reflection path may be available; moreover, the rotation of the
smartphone is typically not properly estimated by sensors, while being crucial for
beamforming. Other approaches has focused on the optimization of sweeping, for
example by introducing a hierarchical search starting from broader and less pene-
trating beams to then use more focused beams. Another solution is based on the
observation that users typically enter the cell from particular directions (correspond-
ing for example to streets), while other directions are not possible due to blockage
effect toward the BS: by properly learning these typical patterns (also maybe related
to commuting habits and changing over the day) and checking more frequently direc-
tions that are typically used for entrance, the duration of the initial access procedure
can be significantly reduced [11].
4 5G Applications
As tracing the channel variations is complicated, the devices should not move fast
while exchanging data over the mm-waves. In this context, two scenarios applications
are particular promising for mm-waves: backhauling and fixed wireless access.
Backhauling refers to the communication among BSs of the same cellular system
and is particularly useful in a mm-wave scenario, where cells are small for coverage
purposes, and connecting all these BS to the fixed network may be expensive. On the
other hand, mm-waves are ideal for backhauling, since they offer a huge spectrum
and the connected devices are fixed, thus not posing channel tracking problems.
Moreover, positions of BSs and antennas can be properly chosen in order to avoid
blockage. For a survey on backhauling for 5G systems see [12]. A typical open issue
in backhauling is the scheduling of transmissions to the backhaul BS and the user
terminals. In this sense the spatial diversity experienced by the users with respect to
the connected BS reduces interference and eases the resource allocation among the
two sets of links.
A second important application of mm-wave for 5G is the fixed wireless access
(FWA), where the wireless connection between the BS and the user terminal replaces
the conventional broadband wired/fiber access at home or in office. Also in this case,
the fixed position of the user is particular suitable for a mm-wave link. Typically the
BS is serving many users (that can also be a mix of FWA users and conventional
moving terminals) and suitable strategies for resources allocation between the two
sets of users must be provided.
4.1 Regulatory Aspects
The ITU coordinates the spectrum allocation worldwide, in particular for the por-
tions of the spectrum that are used for cross-country communications, for astronomy
observations and for satellite observations. The use of the spectrum at a national
level (which is the case for mm-waves within 5G systems) is instead in the hands
of each state. Agreements among states (e.g., within Europe or the United States of
America, USA) may partially or totally delegate the spectrum coordination to inter-
national bodies. In Europe various bodies are involved in the spectrum allocation,
and since 2002 they are coordinated by European Regulators Group for Electronic
Communications Networks and Services. In USA the responsible body is the Fed-
88 Stefano Tomasin
eral Communication Commission (FCC). Once portions of mm-wave spectrum is

allocated to 5G systems, frequency blocks are then leased to operators, typically
through auctions.
In Italy, in October 2018 the auction of groups of frequencies in the range 3600-
3800 MHz and 26.5-27.5 GHz has been completed by the Ministero per lo sviluppo
economico with total selling price of more than 4.5 bilions of euros. In USA an
auction for portions of spectrum around 24 and 28 GHz will be started by FCC in
mid November 2018.
References
1. Rappaport T.S., Xing Y., MacCartney G. R., Molisch A. F., Mellios E., Zhang J.: Overview
of Millimeter Wave Communications for Fifth-Generation (5G) Wireless Networks-with a
focus on Propagation Models. IEEE Transactions on Antennas and Propagation (2017) doi:
10.1109/TAP.2017.2734243
2. 3GPP (2018) Study on channel model for frequencies from 0.5 to 100 GHz. TR
38.901 Ver. 15.0.0 https://portal.3gpp.org/desktopmodules/Specifications/ SpecificationDe-
tails.aspx?specificationId=3173
3. Niu, Y., Li, Y., Jin, and Vasilakos A. V.: A survey of millimeter wave communica-
tions (mmWave) for 5G: opportunities and challenges. Wireless Networks (2015) doi:1
0.1007/s11276-015-0942-z
4. Rappaport T., Sun S., Mayzus R., Zhao H., Azar Y., Wang K., Wong G. N., Schulz J. K.,
Samimi M. and Gutierrez F.: Millimeter wave mobile communications for 5G cellular: it will
work! IEEE Access (2013) doi: 10.1109/ACCESS.2013.2260813
5. Heath R. W., González-Prelcic N., Rangan S., Roh W. and Sayeed A. M.: An overview of
signal processing techniques for millimeter wave MIMO systems. IEEE Journal of Selected
Topics in Signal Processing (2016) doi: 10.1109/JSTSP.2016.2523924
6. Roh W., Seol J., Park J., Lee B., Lee J., Kim Y., Cho J., Cheun K., Aryanfar F.:
Millimeter-wave beamforming as an enabling technology for 5G cellular communications:
theoretical feasibility and prototype results. IEEE Communications Magazine (2014) doi:
10.1109/MCOM.2014.6736750
7. Ahmed I., Khammari H., Shahid A., Musa A., Kim K. S., De Poorter E., and Moerman I.: A
Survey on hybrid beamforming techniques in 5G: architecture and system model perspectives.
IEEE Communications Surveys & Tutorials (2018) doi: 10.1109/COMST.2018.2843719
8. Montagner S., Benvenuto N. and Tomasin S.: Taming the complexity of mm-wave massive
MIMO systems: Efficient channel estimation and beamforming. IEEE International Confer-
ence on Communication Workshop (ICCW)(2015) doi: 10.1109/ICCW.2015.7247349
9. Soleimani H., De Donno D. and Tomasin S.: mm-Wave channel estimation with accelerated
gradient descent algorithms. EURASIP Journal of Wireless Communications and Networks
(2018)
10. Giordani M., Mezzavilla M. and Zorzi M.: Initial access in 5G mmWave cellular networks.
IEEE Communications Magazine (2016) doi: 10.1109/MCOM.2016.1600193CM
11. Soleimani H., Parada R., Tomasin S. and Zorzi M.: Statistical approaches for initial access in
mmwave 5G systems. Proc. European Wireless Conference (2018)
12. Jaber M., Imran M. A., Tafazolli R., Tukmanov A.: 5G backhaul challenges and emerging
research directions: A Survey. IEEE Access (2016) doi: 10.1109/ACCESS.2016.2556011
Massive MIMO
Luca Sanguinetti
Abstract The aim of this chapter is to provide an entry point to Massive MIMO,
as well as an up-to-date survey of the state-of-the-art results in spectral efficiency,
that will guide the evolution of Massive MIMO in the years to come. Particu-
larly, it builds upon the book [1] that is freely accessible in PDF from the website
www.massivemimobook.com.
1 Introduction
The area throughput is highly relevant in contemporary and future cellular networks.
It is measured in bit/s/km2 and determined by three key factors:
B [Hz] × Spectral efficiency [bit/s/Hz/cell]
Area throughput [bit/s/km2 ] = (1)
Cell size [km2 /cell]
where B is the bandwidth, and the spectral efficiency (SE) is the amount of informa-
tion that can be transferred per second over one Hz of bandwidth. Despite inherently
dependent, the three components can be treated as independent as a first-order ap-
proximation. Consequently, the area throughput can be improved by:
1. Allocating more bandwidth;
2. Densifying the network by deploying more base stations (BSs);
3. Improving the SE per cell.

The traditional way to increase the area throughput in cellular networks is to
allocate more bandwidth and to deploy more BSs (i.e., to reduce cell size). Con-
sequently, contemporary networks are already densely deployed and the bandwidth
Prof. L. Sanguinetti, Ph.D. (IEEE Senior Member)

University of Pisa, Dipartimento di Ingegneria dell’Informazione, Italy
e-mail: luca.sanguinetti@unipi.it
89
90 Luca Sanguinetti
left in the sub-6 GHz bands (that are attractive for wide-area coverage) is scarse [2].
In contrast, the growth in SE has been rather modest over the last years.
Nowadays, one of the most promising wireless technology to improve the SE
is Massive MIMO (multiple-input multiple-output) [3, 4, 5]. This physical-layer
technology was introduced by Thomas Marzetta’s seminal paper from 2010 [7]
and equips each BS with an array of many antennas. This allows for coherent
multiuser MIMO transmission where tens of user equipments (UEs) can be served
on each time-frequency resource by spatial multiplexing, in both the uplink (UL)
and downlink (DL) of each cell. The multiplexing gain allows to improve the SE
per cell by orders-of-magnitude [1]. In [7], Marzetta showed that a system with
an extremely large number of BS antennas should operate in time-division duplex
(TDD) mode and exploit channel reciprocity to acquire all the necessary channel
state information (CSI) from a finite number of UL pilot signals. Massive MIMO has
since then gradually changed from a controversial theoretical concept to a mainstream
technology that has found its way into the 5G standard [8].
2 What is Massive MIMO?

Although Massive MIMO is widely considered as one of the key technologies to
provide high SE, it lacks a concise and universal definition. The telecom industry
treats it as multiuser MIMO with many BS antennas, in order to quickly reach the
market. To demonstrate how such a physical-layer technology can improve the SE
by orders-of-magnitude, in this chapter we will consider a stricter definition.
Definition 1 A Massive MIMO cellular network consists of:

• L ≥ 2 cells operating according to a synchronous TDD protocol;
• BSs equipped with M ≥ 64 antennas with fully digital transceiver chains;
• Linear combining and precoding schemes capable of spatially multiplexing

K ≥ 8 UEs per cell;
• More BS antennas than active UEs: M/K > 1.
This definition is in line with the canonical form of Massive MIMO for sub-
6 GHz bands in [1] and includes Marzetta’s setup from [7] as a special case as
M → ∞. It is also in line with real-time Massive MIMO testbeds [9] and field trials
[10]. However, there are important research efforts that deviate from it. Finding an
efficient frequency-division duplexing (FDD) protocol for Massive MIMO is highly
desirable, since there are vast amounts of spectrum reserved for FDD operation.
However, the estimation and feedback overhead of FDD operation in mobile scenarios
is prohibitive, unless something is done to reduce it. The predominant approach is to
Massive MIMO 91
Frequency
Coherence time Tc
Coherence
UL data DL data
τu τd bandwidth
Bc
UL pilots: τp
Time
Fig. 1 The time-frequency plane is divided into coherence blocks over which the channel response
is time-invariant and frequency-flat. The τc samples of each block are used in a TDD fashion for
UL pilots, UL data, and DL data.
assume that there is some kind of channel sparsity that can be utilized to reduce the
channel estimation and feedback overhead. This line of research is quite rich (e.g.,
[11, 12]) but the underlying sparsity hypothesis has not been proved experimentally
at sub-6 Ghz frequencies [13]. Another deviation from the canonical form is the
use of the mmWave band, which refers to the frequency range from 30 − 300 GHz.
Although there is a vast literature that use the ”Massive MIMO” term for both sub-6
GHz and mmWave applications, this can be very confusing because the multiantenna
technology has rather different characteristics in these two applications. Massive
MIMO at sub-6 GHz spectrum can increase the efficiency of highly loaded cells, by
upgrading the technology at existing BSs. In contrast, the huge available bandwidth
in mmWave band can be utilized for high-capacity services, but only over short
distances due to the severe pathloss and high noise power.
2.1 The Transmission Protocol

Massive MIMO can only make efficient use of antennas if each BS knows the
UE channels. For this purpose, a TDD protocol is used where the UL and DL
transmissions fit into the channel coherence block, as illustrated in Fig. 1. This
represents the time-frequency block in which the channels can be approximated
as time-invariant and flat-fading. The number of samples τc in a coherence block
is determined by the coherence time Tc and coherence bandwidth Bc of channels.
Typical values for τc ranges from hundreds (with high-mobility and high-channel
dispersion) to thousands of samples (with low-mobility and low-channel dispersion)
[1, Remark 2.1]. As shown in Fig. 1, the τc samples are used1 for three purposes:
1) τp samples for UL pilots; 2) τu samples for UL reception; 3) τd samples for DL
transmission.
The UL pilots enable the BS to estimate UE channels. Since the TDD protocol is
matched to the coherence block, the UL and DL channels can be considered recip-
rocal2 and the BS can use channel estimates for UL reception and DL transmission.
1 Variations of the TDD protocol above exist, for which we refer to [14, 15].
2 The radio frequency propagation channels are reciprocal by nature, but the end-to-end channels
are also affected by the transceiver hardware. We refer to [17] for reciprocity calibration algorithms.
92 Luca Sanguinetti
f =1 f =2 f =4
Fig. 2 Illustration of a cellular network with a pilot reuse factor of f = 1, 2 or 4. Each group is
indicated with a distinct color and uses a disjunct set of pilots.
A common assumption is τp = K f ≤ τc , so that a pilot book Φ ∈ Cτ p ×τ p of

K f mutually orthogonal pilot sequences can be created, for example, by using the
columns of a Walsh-Hadamard matrix [1]. The pilot book is divided into f groups
with K = τp / f pilots each, where the integer f is called the pilot reuse factor.
Each cell is associated with one of these disjunct groups, according to a predefined
pilot reuse pattern, as illustrated in Fig. 2. Hence, the K UEs in a cell use mutually
orthogonal pilots. The pilot associated with UE k in cell j is denoted by φ jk ∈ Cτ p
and is transmitted with power ρul . This implies φ Tjk φ ∗jk = τp ρul . For simplicity, we
assume that UE k in each cell uses the same pilot, and call P j ⊂ {1, . . . , L} the
group of cells that utilize the same pilot set of cell j.
2.2 Spatially Correlated Channels

j
We denote by hlk ∈ C M the channel between UE k in cell l and BS j. In the Mas-
j j
sive MIMO literature, this channel is often modeled as hlk ∼ NC (0 M , βlk I M ) [3]
where the Gaussian distribution accounts for the random small-scale fading, while
j
βlk describes the macroscopic large-scale fading. This channel model is called un-
correlated Rayleigh fading, and makes the analysis as well as the optimization of
Massive MIMO tractable. However, it is practically questionable. Indeed, propaga-
tion channels are spatially correlated, as seen from measurement campaigns [16] and
by the fact that uncorrelation only appears under extreme physical circumstances.
[1, Sec. 2.2]. A tractable way to model spatially correlated channels is the correlated
Rayleigh fading model:
j j
hlk ∼ NC 0 M , Rlk (2)
j
where Rlk ∈ C M×M is the spatial correlation matrix, assumed to be known at
the BS. As with uncorrelated Rayleigh fading, the Gaussian distribution models
j
the small-scale fading variations, while Rlk describes the macroscopic propagation
j 1 j
characteristics. The normalized trace βlk = M tr(Rlk ) is the average channel gain
from an antenna at BS j to UE k in cell l. Uncorrelated Rayleigh fading with
j j j
Rlk = βlk I M is a special case of this model, but Rlk is in general not diagonal. The
j j
eigenstructure of Rlk determines the spatial channel correlation of hlk ; that is, which
spatial directions are statistically more likely to contain strong signal components
Massive MIMO 93
than others. Strong spatial correlation is characterized by large eigenvalue variations.

j
We refer to [1, Sec. 7.3] for a detailed description of how to generate Rlk .
2.3 Channel Estimation

As said earlier, to make efficient use of its antennas, a given BS j needs to learn
channels. Since the pilots are transmitted synchronously in all cells, BS j can use
them to estimate the channels from its own UEs and the channels from UEs in other
cells. The latter estimates can be used for inter-cell interference suppression.
p
Channel estimates at BS j are obtained from the received pilot signal Y j ∈
C M×τ p , which is given by
K
Õ L
Õ K
Õ
p j j p
Yj = h ji φ Tji + hli φ liT + N j (3)
i=1 l=1,l,j i=1
| {z } | {z } |{z}
Desired pilots Inter-cell pilots Noise
where N j ∈ C M×τ p is thermal noise with i.i.d. elements distributed as NC (0, σul2 ). If
p
j
the statistics are known, the minimum mean-squared error (MMSE) estimator of hli
can be computed as follows.
j
Theorem 1 The MMSE estimate of hli is
1

j j j −1 p
ĥli = Rli Qli √ Y j φ li∗ (4)
τp ρul
p
where Y j is given in (3) and
2
j
Õ
j 1 σul
Qli = Rl0 i + IM . (5)
l0 ∈ Pl
τp ρul
j j j j
The estimation error h̃li = hli − ĥli is independent of ĥli and has correlation matrix
j j j j j j j j −1 j
Cli = E{h̃li (h̃li )H } = Rli − Φli with Φli = Rli (Qli ) Rli .
The normalized mean-squared error (NMSE) given by
E{khli − ĥli k 2 }
j j j j j −1 j
tr Cli tr Rli Qli

Rli
j
= j =1− j
(6)
E{khli k 2 } tr Rli tr Rli

shows that the interference generated by the pilot-sharing UEs in Pl , which enters
j
into Qli in (5), increases it and thus reduces the channel estimation quality. This
“pilot interference” is called pilot contamination and behaves differently from noise;
it not only reduces the estimation quality, but impacts also the SE since (4) contains
j j
the channels of pilot-sharing UEs [1, Sec. 4.4.2]. If Rli Rl0 i = 0 M , the NMSE in
94 Luca Sanguinetti
(6) is completely unaffected by UE k in cell l [1, Sec. 3.3.2]. Therefore, in theory,

pilot contamination between two UEs can be completely avoided if their correlation
matrices are spatially orthogonal. While this condition is unlikely to hold in practice
j j
[1, 5], assigning pilots such that tr(Rli Rl0 i ) is rather small is a good rule-of-thumb
[18].
3 Spectral Efficiency
We analyze the achievable SE, focusing for simplicity only on the UL. The data signal
from UE k in cell j is denoted by s jk ∼ NC (0, ρul ), with ρul being the transmit power.
To detect s jk , BS j selects the combining vector v jk ∈ C M , which is multiplied with
the received signal y j ∈ C M at BS j. This yields
K
Õ L
Õ K
Õ
j j j
vHjk y j = vHjk h jk s jk + vHjk h ji s ji + vHjk hli sli + vHjk n j (7)
i=1,i,k l=1,l,j i=1
| {z } | {z } |{z}
Intra-cell interference Inter-cell interference Noise
where n j ∼ NC (0 M , σul2 I M ) is independent noise. An achievable SE is any number

that is below the capacity. While the classical “Shannon formula” cannot be applied
when the receiver has imperfect CSI, there exist well-established capacity lower
bounds that can be used [1, 3].
Theorem 2 If the MMSE channel estimation is used, an UL SE of UE k in cell j is
τu n o
SEul = E log 2 1 + γ ul
[bit/s/Hz] (8)
jk
τc jk
with τu /τc accounting for the fraction of samples used for UL data and
|vHjk ĥ jk | 2
j
γ ul
jk = ! (9)
L K K
j j H j j H
H
+ + Z j v jk
Í Í Í
v jk ĥli (ĥli ) ĥ jk (ĥ jk )
l=1,l,j i=1 i=1,i,k
2
σul
j j
where Z j = Rli − Φli +
Í L ÍK
l=1 i=1 ρul I M .
The bound in (8) has been analyzed in a number of articles that consider heuristic
linear detectors [3]. Two popular choices for v jk are maximum-ratio (MR) and
zero-forcing (ZF) combining:
 VMR bj
=H with MR combining


j j
V j , v j1 . . . v jK = (10)

−1
j H
 VZF
 bj
=H H
b H bj with ZF combining
 j j j j
with Hb j = [b j
h j1 . . . b
j
h jK ] ∈ C M×K containing the estimates of intra-cell channels
j
in cell j. In a single-cell scenario with perfect CSI, MR and ZF are asymptotically
Massive MIMO 95
Table 1 Network parameters for SE evaluation. The asymmetric network (with wrap-around) in
Fig. 2 is considered.
Parameter Value
Cell area 1 km × 1 km
Number of cells and UEs per cell L = 16, K = 10
UL noise power and UL transmit power σul2 = −94 dBm, ρul = 20 dBm
Samples per coherence block τc = 200
Pilot reuse factor f = 1, 2 or 4
j
Distance between UE k in cell l and BS j dl k
Large-scale fading coefficient for
j
j dl k j
βl k = −148.1 − 37.6 log10 1 km + Fl k dB
the channel between UE k in cell l and BS j
j
Shadow fading between UE k in cell l and BS j Fl k ∼ N(0, 10)
optimal at low and high SNRs [1, Sec. 4.1.1]. However, in a multicellular network
with imperfect CSI and pilot contamination, they are both suboptimal. Instead of
resorting to heuristics, we notice that the SINR in (8) has the form of a generalized
Rayleigh quotient. Hence, the maximum is achieved by [5]:
L
!−1
VM−MMSE bj H
Õ
, v j1 . . . v jK = bj + Zj bj. (11)

j Hl
Hl
H j
l=1
This optimal combining scheme also minimizes the conditional MSE E{|s jk −
vHjk y j | 2 | {H
b j }}. It was introduced in [5, 19] and called multicell MMSE (M-MMSE)
l
combining. The “multicell” notion was used to differentiate it from the single-cell
MMSE (S-MMSE) combining scheme [20], which is widely used in the Massive
MIMO literature and given by:
j j H −1
VS−MMSE , v j1 . . . v jK = H + Zj H bj (12)

b H b
j j j j
2
σul
j j j
with Z j = i=1 R ji − Φ ji + l=1,l,j i=1 Rli + ρul I M . The main difference from
ÍK ÍL ÍK
(11) is that only channel estimates in the own cell are computed and utilized in
S-MMSE, while ĥ jli ĥHjli − Φ jli is replaced with its average (i.e., zero) for all UEs in
other cells.
3.1 Performance Evaluation

To quantify the SE that can be achieved in Massive MIMO, we now consider the
network setup in Fig. 2 with the parameters given in Table 1. Each BS is equipped
with a uniform linear array with half-wavelength antenna spacing. Each channel
consists of S = 6 scattering clusters, which are modeled by the Gaussian local
j
scattering model [1, Sec. 2.6]. Hence, the (m1, m2 )th element of Rli is given by
96 Luca Sanguinetti
70
M-MMSE
Average sum SE [bit/s/Hz/cell]

60 S-MMSE
ZF
50 MR
40
30
20
10
0
10 32 64 100 150 200 250
Number of antennas (M)
Fig. 3 Average UL sum SE as a function of M with different combining schemes with the correlated
Rayleigh fading model in (13), by using the network setup in Fig. 2 with pilot reuse factor f = 1.
S 2 2
h
j
i
j
fm1 + fm2 1 Õ jπ(m1 −m2 ) sin(ϕlji, s ) − σ2ϕ π(m1 −m2 ) cos(ϕlji, s )
Rli = βli 10 10 e e (13)
m1,m2 S s=1
where βli is reported in Table 1 and fm ∼ N (0, σf2 ) represents i.i.d. log-normal
j
channel gain variations with σf = 2, which model the gain variations observed from
j
measurements in [13]. Let ϕli be the geographical angle to UE i in cell l as seen from
BS j. Cluster s is characterized by the randomly generated nominal angle-of-arrival
j j j
ϕli,s ∼ U[ϕli − 40◦, ϕli + 40◦ ] and the angles of the multipath components are
Gaussian distributed around the nominal angle with standard deviation σϕ2 = 5◦ .
Fig. 3 shows the average UL sum SE for the pilot reuse factor f = 1. M-MMSE
provides the highest SE, which passes from 14.89 bit/s/Hz to 62.47 bit/s/Hz as M
increases. The suboptimal schemes are quite competitive when M is small, but in
the Massive MIMO regime of M ≥ 64, the losses are noticeable. The superior
SE of M-MMSE for any value of M comes from the fact that it finds the optimal
tradeoff between interference suppression and coherent combining of the desired
signal. When compared with the UL sum SE 2.8 bit/s/Hz/cell achieved by basic
LTE systems, the SE is increased by more than 10× with ZF when M ≥ 64, which
increases to 15× with M-MMSE. This provides evidence that the Massive MIMO
technology is capable of improving the SE by an order of magnitude.
Table 2 reports the average UL sum SE for M = 100 and different pilot reuse
factors. M-MMSE benefits particularly much from having f > 1. Thanks to the
improved channel estimation quality, M-MMSE can better suppress the interference
from UEs in the surrounding cells and the SE gain is increased. Since the other
schemes do not suppress interference from other cells, their SE reduces when f is
increased.
Massive MIMO 97
Table 2 Average UL sum SE for M = 100 and different pilot reuse factors f .
f =1 f =2 f =4
M-MMSE 48.71 54.91 56.09

S-MMSE 43.69 43.81 41.36
ZF 39.26 39.82 36.91
MR 18.94 18.42 16.55
4 Pilot contamination is not a fundamental asymptotic limitation

Marzetta showed in his seminal paper [7] that, under spatially uncorrelated channels,
MR combining (and precoding) achieves a finite asymptotic SE as M → ∞ whose
value is only determined by pilot contamination. The same limit is achieved (but for
a smaller number of antennas) by S-MMSE and ZF [20]. Since then, the following
results about Massive MIMO had been taken for granted:
• Due to pilot contamination, the capacity saturates as M → ∞;
• MR is asymptotically optimal;
• More sophisticated schemes than MR can only improve the SE for finite M.
Recently, [5, 6] showed that with M-MMSE and a tiny amount of spatial channel
correlation, the capacity of Massive MIMO increases without bound in UL and DL
as M → ∞, even under pilot contamination. More precisely, [5] showed that, if the
BS makes use of channel estimates from UEs in all cells, an unbounded capacity is
achieved with Massive MIMO when the channel correlation matrices of the pilot-
contaminating UEs are asymptotically linearly independent. This is generally the
case in practice [16]. If also the diagonals of the correlation matrices are linearly
independent, [5, 6] proved that it is sufficient to know these diagonals (not the full
correlation matrices) to achieve an unbounded asymptotic SE. Similar results were
shown in [21] for a generalized MR combining.
All this proves that the above results are not correct, and tells us how to design
Massive MIMO in the years to come. Indeed, the purpose of analyzing the asymp-
totic SE when M → ∞ is not to advocate BSs with a nearly infinite number of
antennas—that is physically impossible in a finite-sized world and the conventional
channel models will eventually break down since more power is received than was
transmitted. The importance of asymptotics is instead what it tells us about practical
networks as antennas become a commodity and are deployed everywhere. For ex-
ample, consider a network with any finite number of UL pilot signals and of active
UEs, each with a finite-valued SE requirement. The results in [5] imply that there
exists a finite number of antennas, M, that allows to deliver any required SE even in
the presence of pilot contamination. This is made possible by exploiting the spatial
correlation that appears naturally in wireless channels. The lack of these insights has
not prevented the first deployments of Massive MIMO, but will guide the evolution
of the technology towards what we call Massive MIMO 2.0.
98 Luca Sanguinetti
References
1. E. Björnson, J. Hoydis, and L. Sanguinetti (2017) “Massive MIMO networks: Spectral, energy,
and hardware efficiency” Foundations and Trends in Signal Processing 3-4: 154 - 655.
2. J. G. Andrews and X. Zhang and G. D. Durgin and A. K. Gupta (2016) “Are we approaching
the fundamental limits of wireless network densification?” IEEE Commun. Mag. 54:184 -
190.
3. T. L. Marzetta, E. G. Larsson, H. Yang, and H. Q. Ngo, (2016) “Fundamentals of Massive
MIMO” Cambridge University Press.
4. E. G. Larsson, F. Tufvesson, O. Edfors, and T. L. Marzetta (2014) “Massive MIMO for next
generation wireless systems” IEEE Commun. Mag. 2: 186 - 195.
5. E. Björnson, J. Hoydis, and L. Sanguinetti (2018) “Massive MIMO has unlimited capacity”
IEEE Trans. Wireless Commun. 1: 574 - 590.
6. E. Björnson, J. Hoydis, and L. Sanguinetti (2017) “Pilot contamination is not a fundamental
asymptotic limitation in Massive MIMO” IEEE International Conference on Communications
(ICC) 2017.
7. T. L. Marzetta (2010) “Noncooperative cellular wireless with unlimited numbers of base
station antennas” IEEE Trans. Wireless Commun. 11: 3590 - 3600.
8. S. Parkvall and E. Dahlman and A. Furuskär and M. Frenne (2017) “NR: The New 5G Radio
Access Technology” IEEE Commun. Std. Mag. 4: 24 - 30.
9. S. Malkowsky, J. Vieira, L. Liu, P. Harris, K. Nieman, N. Kundargi, I. C. Wong, F. Tufvesson,
V. Owall, and O. Edfors (2017) “The worlds first real-time testbed for Massive MIMO: Design,
implementation, and validation” IEEE Access 5: 9073 - 9088.
10. G. Liu, X. Hou, J. Jin, F. Wang, Q. Wang, Y. Hao, Y. Huang, X. Wang, X. Xiao, and A. Deng
(2017) “3D-MIMO with massive antennas paves the way to 5G enhanced mobile broadband:
From system design to field trials” IEEE J. Sel. Areas Commun. 6: 1222 - 1233.
11. A. Adhikary, J. Nam, J.-Y. Ahn, and G. Caire (2013) “Joint spatial division and multiplexing
- the large-scale array regime” IEEE Trans. Inf. Theory 10: 6441 - 6463.
12. J. Choi, D. Love, and P. Bidigare (2014) “Downlink training techniques for FDD massive
MIMO systems: Open-loop and closed-loop training with memory” IEEE J. Sel. Topics
Signal Process. 5: 802 - 814.
13. X. Gao and O. Edfors and F. Rusek and F. Tufvesson (2015) “Massive MIMO Performance
Evaluation Based on Measured Propagation Data” IEEE Trans. Wireless Commun. 7: 3899 -
3911.
14. K. Upadhya, S. A. Vorobyov, and M. Vehkapera (2017) “Superimposed pilots are superior for
mitigating pilot contamination in Massive MIMO’ IEEE Trans. Signal Process. 11: 2917 -
2932.
15. D. Verenzuela, E. Björnson,and L. Sanguinetti (2018) “Spectral and energy efficiency of
superimposed pilots in uplink Massive MIMO” IEEE Trans. Wireless Commun. 11: 7099 -
7115.
16. X. Gao and O. Edfors and F. Tufvesson and E. G. Larsson (2015) “Massive MIMO in Real
Propagation Environments: Do All Antennas Contribute Equally?” IEEE Trans. Commun. 11:
3917 - 3928.
17. J. Vieira and F. Rusek and O. Edfors and S. Malkowsky and L. Liu and F. Tufvesson (2017)
“Reciprocity Calibration for Massive MIMO: Proposal, Modeling, and Validation” IEEE
Trans. Wireless Commun. 5: 3042 - 3056.
18. H. Yin, D. Gesbert, M. Filippou, and Y. Liu (2013) “A coordinated approach to channel
estimation in large-scale multiple-antenna systems” IEEE J. Sel. Areas Commun. 2: 264 - 273.
19. Xueru Li, Emil Björnson, Erik G. Larsson, Shidong Zhou, Jing Wang (2017) “Massive
MIMO with Multi-cell MMSE Processing: Exploiting All Pilots for Interference Suppression”
EURASIP J. Wireless Commun. Networking 117.
20. J. Hoydis, S. ten Brink, and M. Debbah (2013) “Massive MIMO in the UL/DL of cellular
networks: How many antennas do we need?” IEEE J. Sel. Areas Commun. 2: 160 - 171.
21. D. Neumann; T. Wiese, M. Joham, W. Utschick (2018) “A Bilinear Equalizer for Massive
MIMO Systems” IEEE Trans. Signal Process. 14: 3740 - 3751.
Overview on Optical Fronthauling Technologies
for Fixed-Mobile Convergence
Roberto Gaudino
Abstract We present in this Chapter an overview of optical fronthauling technologies

for fixed-mobile convergence in next generation 5G access networks. The Chapter
first introduces the general concept of Cloud Radio Access Network, or C-RAN, then
it presents the most common proposals for fronthauling, based on Digitized Radio-
over-Fiber, or D-RoF, according to the CPRI or OBSAI standards. The Chapter then
prosecutes by presenting the very recent evolutions of D-RoF toward the "functional
split" paradigm, as already available in the latest releases of the CPRI specifications.
Finally, some recent research trends towards Analog RoF are presented. The Chapter
is intentionally written in a tutorial way, to be used by newcomers in this field. It
anyway also reports a vast set of bibliographic references to guide the interested
reader toward more detailed technical presentations.
1 Introduction and scope of this Chapter
In modern mobile access network (4G and in the forthcoming 5G), telco operators are
trying to reduce their network CAPEX and OPEX implementing the new paradigm
usually indicated as Cloud- (or Centralized-) Radio Access Network (C-RAN). In a
broad sense, C-RAN is an architecture in which several physical and network layer
functions that were previously implemented in base-stations (BS) are moved (and
thus centralized) to Central Offices (CO). Advantages of this approach are described
in many recent papers, such as [1], and can be summarized as follows:
• Reduction of the complexity of the antenna-site hardware and software. In partic-
ular, the antenna site installation is potentially simplified, footprint is lower and
the maintenance costs are largely reduced.
Roberto Gaudino
Politecnico di Torino, Dipartimento di Elettronica e Telecomunicazioni (DET), Corso Duca degli
Abruzzi, 24, 10129 Torino, Italy, e-mail: roberto.gaudino@polito.it
99
100 Roberto Gaudino
• Better energy efficiency: centralization and function virtualization allow to dy-

namically allocate centralized processing capabilities, and depending on traffic
requirements, processing can be turned to low power or even be shut down
selectively, thanks to the usual advantages arising from resources statistically
multiplexing.
• Multi-point cooperation techniques proposed for 5G can be much more efficiently
implemented thanks to the centralized processing of physical and network layer
functions for several antenna sites.
The C-RAN paradigm requires anyway much larger transfer of information from
the CO to the antenna sites compared to previous solutions based on back-hauling,
thus requiring dedicated high-capacity links in the fixed access networks that has to
support the mobile network. The techniques used to implement these links in C-RAN
is usually indicated as "fronthauling", and typically requires a fixed access networks
based on optical fibers. The trend is often indicated as "fixed-mobile convergence",
and it is enabled by a joint design of the fiber-based fixed access network and the
mobile network.
Scope of this Chapter is to give an introductory overview of current trends in
optical fronthauling. The Chapter is intentionally written in a tutorial way, and it
is thus meant for people that are new in the sector of optical fronthauling and who
wants to have a first insight. After reading this Chapter, the interested reader can find
much more detailed technical information in the large list of papers reported in the
Bibliography [1]-[24].
The Chapter is thus organized as follows. We start by presenting in the rest of this
Section some preliminary concepts on fronthauling, then we focus in the following
Sections on the different fronthauling techniques that are currently implemented or
under investigation, and in particular:
• Digitized Radio-over-Fiber (D-RoF), based on the CPRI [6] or OBSAI [7] de-facto
standards.
• The "network level" trends, based on the so-called "functional split" approach,
following the recently released e-CPRI [8] specifications.
• Alternative solutions based on Analog Radio-over-Fiber (A-RoF) or DSP-assisted
equivalents of A-RoF
In order to have a common terminology, we introduce here some of the basic
definitions and acronyms following the conventions used, for instance, in [1], [2]
and [3], also reporting them in Fig. 1, which shows a schematic comparison between
traditional architectures based on backhauling and the new C-RAN+fronthauling
architectures. The key elements of the C-RAN+fronthauling architecture are:
• the remote radio head (RRH) is the part of the hardware and software functions
that remains in the antenna site. The C-RAN target is to leave in the RRH only
the radio frequency (RF) part and a very lean protocol for interfacing it to the
fronthauling link.
• most of the baseband digital signal processing functions that are usually present
in the "traditional" base station architecture are moved to Base-Band processing
Units (BBU) at the CO.
Overview on Optical Fronthauling Technologies for Fixed-Mobile Convergence 101
• in "true" Cloud RAN, BBU are organized in a shared BBU pool in (one or more)
central offices, to allow virtualisation of the functions and to take advantage of
statistical multiplexing arising from the centralization of the services for many
BBUs.
Fig. 1 Comparison with traditional and C-RAN architectures for mobile networks.
2 Fronthauling based on Digitized Radio-over-Fiber D-RoF
The most common fronthauling solution, that has already reached commercial level
implementation, is based on the Digitized Radio-over-Fiber (D-RoF) approach. Here
we will report its main features, while the interesting reader can find more details
in the Specification of one of the two available de-facto standards written by two
different industry consortia, called CPRI [6] (Common Public Radio Interface) and
OBSAI [7] (Open Base Station Architecture Initiative). In Fig. 2 we present some
simplified schematics that try to point out the main differences between traditional
architectures (which we will simply indicated as "backhauling") and new DRoF
fronthauling. The top left graph shows the traditional backhauling architecture, where
the link between the base station and the CO basically carries the same packets that
flows on the wireless part (plus obviously some additional control and management
information). The top right graphs zooms on the RF and BBU parts, while the bottom
graph shows the DRoF fronthauling architecture. The basic principle of D-RoF is
transporting over the fronthauling fiber links the "native" radio signal, such as in
analog radio-over-fiber, but using a digitized approach of the baseband version of the
signal. In particular, as schematized in Fig. 2 and following for instance the upstream
direction:
• the radio signal received from the antenna is processed in the RRH by RF I/Q
demodulation hardware, down-converting it to two I- and Q- baseband signals.
102 Roberto Gaudino
For what we need to discuss later on, it is worth noting that if the RF signal
occupies a bandwidth B in the wireless spectrum, the two I- and Q- baseband
signals occupies a (one-sided) bandwidth B/2 each.
• the two baseband signals are digitized by a pair of analog-to-digital converters
(ADC), generating a digital stream at their output. If the number of bits of the
ADC is nbit and the ADC sampling rate is fs , the resulting bit rate at the output
of one ADC is nbit · fs , so that the total resulting bit rate to be transported on the
fronthauling link is 2nbit · fs , where for the sampling theorem fs > B.
• a very lean protocol (such as CPRI or OBSAI) then adds the required control and
framing information to this digital stream (thus further increasing the bit rate),
then sent it using a suitable optical fiber transport system to the central office,
where it is optically received, processed and sent to the aforementioned BBU.
• overall, as again shown in Fig. 2, the BB functions are completely moved from the
antenna site to the CO, thus allowing to fully implement the C-RAN paradigm.
Fig. 2 Comparison between traditional backhauling and new DRoF fronthauling: simplified
schematics. Top left: traditional backhauling architecture. Top right: zooming in the RF and BBU
parts of the traditional architecture. Bottom: DRoF fronthauling architecture
Compared to the traditional backhauling architecture, the DRoF one poses anyway
two main constraints on the fronthauling link: bit rate to be transported and latency.
Starting from the first issue, let’s consider, as already mentioned, that for a bandwidth
B of a give radio waveform in the wireless spectrum, a bit rate (at least) equal to
2nbit · fs (plus control information) should be carried on the fronthauling link, and
for the sampling theorem fs > B, so that the resulting bit rate must be greater than
2nbit · B. As a more practical example, for B = 20 MHz on the wireless channel, the
CPRI protocol [6] assumes that the DAC/ADC runs at fs =30.72 Msamples/s and
nbit = 16 and thus the net bit rate to be transported is 983 Mbit/s. On top of this,
CPRI add some control and management information, and the 8B/10B line code,
resulting in a gross bit rate to be transported of 1.228 Gbit/s (CPRI line bit rate
option 2). Considering that on a B = 20 MHz wireless bandwidth the net bit rate is
often less than 100 Mbit/s, it is evident that DRoF fronthauling faces the so-called
"bandwidth expansion" issue, i.e., it requires transporting a bit rate that is, as a rule
of thumb, at least 10 times bigger than the net bit rate of the radio part, and thus
also (again as a rule of thumb) 10 times bigger than for the traditional backhauling
architecture.
The bit rate we have estimated is for the fronthauling transport of one single radio
waveform, so that in practice for a given fronthauling link supporting one RRH it
should be multiplied by:
• the number of segments of the antenna site (typically at least three).
• if carrier aggregation is used, the number of radio carriers used by the antenna
site.
• if MIMO techniques are used, the multiplicity of the implemented NxN MIMO.
The resulting high bit rates thus obviously require a fiber-based solution for the
fronthauling link. Using today optical access solutions, the fronthauling bit rates are
typically not critical for 4G LTE networks, where typical parameters are B = 5 or
10 MHz (or less), 3 segments and limited or no MIMO. In many currently installed
implementations for 4G, the optical links thus runs at about 10 Gbit/s (CPRI line
bit rate option 7) or less. The latest version of the CPRI specs introduced CPRI line
bit rate option 10, having a bit rate equal to 24.33 Gbit/s (thus able for instance to
transport in parallel 24 radio waveforms of the type presented in the aforementioned
example with B = 20 MHz). Anyway, the required bit rate may become extremely
critical for LTE-Advanced and even more for 5G, which in some implementations will
use much larger radio bands B, massive MIMO and extensive carrier aggregation.
This is actually the main reason why fronthauling solutions have recently evolved
in the directions described in the following Sect. 3, which requires much lower bit
rates, or in Sect. 4, which follow an analog radio-over-fiber approach. Some research
papers have also tried to address the issue of using digital compression techniques
directly to the bit stream at the output of CPRI, such as [9], which demonstrates a
bit rate compression by a factor of two, and [10] showing a compression by a factor
of about four, at the expense of higher DSP complexity.
Another very important issue in fronthauling compared to traditional backhauling
is that the end-to-end latency should be kept very small. The actual requirements
are strongly dependent on the physical layer specification of the radio part, but it
is evident that while in traditional backhauling architecture the processing is inside
104 Roberto Gaudino
each base station, and thus the latency is a "local" issue inside each given radio
cell, in C-RAN with DRoF fronthauling the round-trip latency is actually the sum
of the two parts due to the wireless and the optical segments. The interested reader
can find detailed information in [2], [4]. Here we just want to point out that the
actual round-trip latency constraints often require that the fronthauling segment has
a total round-trip latency below 200 µs, which should include all processing (such
as CPRI) and propagation (fixed or variable) delays. This tight latency requirement
for fronthauling has several consequences:
• the optical physical layer that supports fronthauling should be kept simple to avoid
adding an additional term to the latency budget. In fact, many practical CPRI
implementation uses pure optical On-Off Keying (OOK) and direct detection,
usually without forward error correcting codes (FEC).
• if DRoF digital stream is to be carried on a packet-switched optical access so-
lutions, then the variable delay that is intrinsic in a multiple access shared envi-
ronment should be kept under strict control. This is for instance very important
for DRoF over Passive OPtical Networks (PON) or in general for any switched
Ethernet transport. A very vast and recent literature is available on this topic, such
as [11] and [12]
• ultimately, the fiber propagation delay limits the maximum geographical extension
of a given C-RAN domain. As an example, a 10 km distance between RRF and
CO accounts for about 100 µs round-trip delay due to fiber propagation alone.
3 Alternative solution based on the functional split approach
As outlined in the previous Section, the CPRI (or OBSAI) DRoF enables the full
implementation of the C-RAN paradigm, at the expense anyway of very high bit rate
requirements on the optical links. Particularly for future 5G millimiter-wave bands,
which may use several hundred MHz radio bands and massive MIMO, the "pure"
CPRI approach becomes unsustainable. As a consequence, several new options
have been proposed mostly in order to greatly reduce the required bit rate on the
fronthauling link. The most promising one is likely the "functional split" paradigm,
a solution that has already been specified by the CPRI industry cooperation under
the acronym eCPRI. While the details of the implementation can be found in [8],
we report here only its key principle, that is based on considering that, rather than
moving the full baseband protocol stack from the RRH to the BBUs, one can envision
moving only a part of it, by properly splitting the functions that should remain in
the RRH, and those that are moved to the BBUs (from which the definition of
"functional split"). In particular, eCPRI decomposes the baseband protocol stack
into the following layers (using the E-UTRA terminology and starting from the
higher layer and going down to the physical layer):
• Radio Resource Control (RRC)
• Packet Data Convergence Protocol (PDCP)
• Radio Link Control (RLC)

• Medium Access Control (MAC)
• Physical (PHY) which is the lower layer that is relevant for C-RAN and that
directly interfaces to the Radio Functions layer (RF)
The eCPRI recommendation indicates five possible functional splits (labelled as
"Split A" to "Split E") that are logically placed "below" any of the aforementioned
layers. For instance, "Split E" is logically implemented below the PHY layer and
coincides with the CPRI approach described in the previous section, in which vir-
tually the full stack is moved in the centralized BBU, while for instance "Split C"
leaves the PHY and MAC function in the RRH and moves only the RLC, PDPC and
RRC layers in the centralized BBUs. The higher the split is placed, the smaller is the
bit rate to be transported. An interesting example is given in [13]: for a situation in
which Split E requires 9.83 Gbit/s, Split D requires 2.68 Gbit/s and Split C requires
468 Mbit/s. The availability of different levels of splitting thus allows the network
designer a trade-off between fronthauling bit rate and the level of C-RAN actual
centralization of functions.
4 Alternative solutions based on variants of Analog

Radio-over-Fiber
While the functional split paradigm tackles the DRoF bandwidth expansion problem
with a "network layer" approach, other solutions have been proposed at the research
level focusing only on the physical layer, using variants of analog radio-over-fiber
(A-RoF) technologies [14],[15] in which the radio waveform is directly carried on
the fiber in analog way according to one of the following principles:
• direct transmission on the fiber of the radio waveform at its original RF fre-
quency, without any down-conversion (sometimes indicated as "RF-over-Fiber"
[15]). This the most traditional analog radio over fiber approach, and it has been
commercially used for antenna remoting or distributed antenna systems (DAS) in
the so-called "Microwave Applications" [17].
• transmission on the fiber of a down-converted version of the radio waveform
at an intermediate frequency, sometimes indicated as "IF-over-Fiber" [15]. This
approach is usually preferred to RF-over-Fiber when the goal is the aggregation of
many radio waveforms on the same fiber using Frequency Division Multiplexing
(FDM) [18].
• more exotic solutions, such as those based on delta-sigma analog-to-digital con-
version, [16] which are anyway for the moment only limited to research level.
Of the three approaches presented in this section, the one originally proposed in
[18] seems particularly suitable for the requirement of next-generation fronthauling
for 5G. It is an IF-over-Fiber approach that allows multiplexing a high number of
radio channels using a digital signal processing (DSP) and FDM aggregation, and it
is often indicated as DSP-assisted A-RoF. The key idea is that, using a proper DSP
106 Roberto Gaudino
setup that uses some of the properties of FFT/IFFT algorithms, a very large number
of baseband radio waveforms can be multiplexed with a single FFT/IFFT operation
on very tightly packed comb of intermediate frequencies. For instance, the original
paper [18] experimentally showed a DSP-assisted FDM aggregation of 48 20-MHz
LTE radio waveforms using only 1.5 GHz analog bandwidth on the fiber, while the
CPRI approach would have required an aggregated data rate of 59 Gb/s.
Our group has worked in the same area, demonstrating:
• the extension of the capacity of these systems up to 384 20-MHz LTE radio
waveforms on the same fiber [19].
• the possibility to carry these signals also on Passive Optical Network (PON)
architectures [20], which are intrinsically very demanding in terms of end-to-end
attenuation of the optical link.
• the adaptation to the specific requirements of upstream transmission [21].
5 Discussion on the relation with optical access networks and

conclusion
We discuss here, as a conclusion, the relation fronthauling and the existing optical
access technologies, also because this is an area that is currently (2018) going
through an enormous revolution. In fact, in most of the developed countries, and
in particular in Europe, fixed access networks are undergoing an epochal transition
from the previous "all-copper" (twisted pairs+ADSL) situation to the new Fiber-To-
The-Home (FTTH) paradigm, as also requested by one of the pillars of the H2020
EU Digital Agenda. As of 2018, and focusing on Italy as an example, the major telco
operators are massively deploying either FTTH or intermediate solutions, such as
Fiber-to-the-Cab (FTTCab) or Fiber-to-the-Building (FTTB), typically on a market
competitive business model in the large and denser cities, and on public incentives
for smaller cities and rural areas.
FTTH deployment for residential users are typically based on PON i.e. on optical-
splitter based point-to-multipoint architectures, while FTTCab is more likely im-
plemented using dedicated point-to-point (P2P) fibers. The fronthauling solutions
presented in this Chapter, and particularly CPRI DRoF, are today usually deployed
using P2P dedicated fibers, since in this case basically no sharing issues should
be handled at the network layer, and at the physical layers the optical loss is low,
making the transceivers particularly simple and thus low cost. Anyway, the massive
deployment of FTTH is making geographicaly widespread PON access more and
more common, at least in urban areas. A very vast scientific literature is thus focused
on enabling efficient fronthauling over PON [11], [12], [18], [19], [20], [21], [22],
[23]. Here, two main technical issues should be solved:
• at the optical physical layer, PON are characterized by a high insertion loss (above
29 dB), mostly due to the presence of optical splitters in the link. The solutions
presented in Sect. 4 are particularly sensitive to high insertion loss, since they use
"analog" transmission technologies. Anyway, by a careful design of the link, many

demonstrations of A-RoF over PON have been demonstrated in the literature [18],
[19].
• at the network layer, PON is intrinsically a shared solution and thus requires
multiplexing. The currently most deployed standard (ITU-T G-PON) use Time
Division Multiple Access (TDMA) in both directions and a Dynamic Bandwidth
Allocation (DBA) algorithm [24] for sharing resources among the (up to 64)
PON users. All the following PON standards that have followed G-PON still
uses some form of TDMA. The use of DRoF (such as CPRI) over PON thus
poses stringent latency requirements. In fact, while the original CPRI idea was
based on a dedicated circuit-switched point-to-point solution, its application over
PON should intrinsically handle the latency variation typical of a packet-switched
architecture. Again, a very vast literature is available on this topic, with several
proposed solutions [11], [12].
We have tried to give in this Chapter an overview of different possible solutions
for optical fronthauling, presenting a tutorial on its main implementation aspects. A
vast literature exists also on the related techno-economics [22], [23].
Acknowledgements The work presented in this Chapter was supported by the POLITO PhotoNext
Center (www.photonext.polito.it). The author would like to thank his collaborators Pablo Torres
Ferrera and Mengesha Befekadu Debebe.
References
1. J. Wu, Z. Zhang, Y. Hong and Y. Wen, "Cloud radio access network (C-RAN): a primer,"
IEEE Network, vol. 29, no. 1, pp. 35-41, Jan.-Feb. 2015.
2. A. Pizzinat, P. Chanclou, F. Saliou and T. Diallo, "Things You Should Know About Fronthaul,"
IEEE/OSA Journal of Lightwave Technology (JLT), vol. 33, no. 5, pp. 1077-1083,March 2015.
3. A. Checko et al., "Cloud RAN for Mobile Networks — A Technology Overview," IEEE
Communications Surveys & Tutorials, vol. 17, no. 1, pp. 405-426, 2015.
4. M. Agiwal, A. Roy and N. Saxena, "Next Generation 5G Wireless Networks: A Comprehensive
Survey," IEEE Communications Surveys & Tutorials, vol. 18, no. 3, pp. 1617-1655, 2016.
5. T. Pfeiffer, "Next generation mobile fronthaul and midhaul architectures," IEEE/OSA Journal
of Optical Communications and Networking, vol. 7, no. 11, pp. B38-B45, 1 November 2015.
6. Official public web page for CPRI specification http://www.cpri.info/spec.html
7. Official public web page for OBSAI specification http://www.obsai.com/specifications.htm
8. Official public web page for e-CPRI specification
http://www.cpri.info/downloads/eCPRI_v_1_2_2018_06_25.pdf
9. B. Guo, W. Cao, A. Tao and D. Samardzija, "LTE/LTE-A signal compression on the CPRI
interface," Bell Labs Technical Journal, vol. 18, no. 2, pp. 117-133, Sept. 2013.
10. H. Si, B. L. Ng, M. S. Rahman and J. Zhang, "A Novel and Efficient Vector Quantization
Based CPRI Compression Algorithm," IEEE Transactions on Vehicular Technology, vol. 66,
no. 8, pp. 7061-7071, Aug. 2017.
11. D. Chitimalla, K. Kondepu, L. Valcarenghi, M. Tornatore and B. Mukherjee, "5G fronthaul-
latency and jitter studies of CPRI over Ethernet," IEEE/OSA Journal of Optical Communica-
tions and Networking (JOCN), vol. 9, no. 2, pp. 172-182, Feb. 2017.
108 Roberto Gaudino
12. L. Valcarenghi, K. Kondepu and P. Castoldi, "Time-versus size-based CPRI in ethernet encap-
sulation for next generation reconfigurable fronthaul," IEEE/OSA Journal of Optical Commu-
nications and Networking (JOCN), vol. 9, no. 9, pp. D64-D73, Sept. 2017.
13. R. Knopp, "Overview of Functional Splits in OAI"
https://www.openairinterface.org/docs/workshop/5_OAI_Workshop_20180620/KNOPP_OAI-
functional-splits.pdf
14. J. Beas, G. Castanon, I. Aldaya, A. Aragon-Zavala and G. Campuzano, "Millimeter-Wave
Frequency Radio over Fiber Systems: A Survey," IEEE Communications Surveys & Tutorials,
vol. 15, no. 4, pp. 1593-1619.
15. C. Lim et al., "Fiber-Wireless Networks and Subsystem Technologies," IEEE/OSA Journal of
Lightwave Technology, vol. 28, no. 4, pp. 390-405, Feb.15, 2010.
16. J. Wang et al., "Digital mobile fronthaul based on delta-sigma modulation for 32 LTE car-
rier aggregation and FBMC signals," IEEE/OSA Journal of Optical Communications and
Networking, vol. 9, no. 2, pp. A233-A244, 2017.
17. J. Yao, "Microwave Photonics," IEEE/OSA Journal of Lightwave Technology, vol. 27, no. 3,
pp. 314-335, Feb.1, 2009.
18. X. Liu, H. Zeng, N. Chand and F. Effenberger, "Efficient Mobile Fronthaul via DSP-Based
Channel Aggregation," IEEE/OSA Journal of Lightwave Technology, vol. 34, no. 6, pp. 1556-
1564, 2015.
19. M. Befekadu, S. Straullu, S. Abrate and R. Gaudino, "Experimental Optimization of DSP-
Aggregated Front-hauling Transmission for up to 4x96 LTE radio waveforms," 42nd European
Conference on Optical Communication, ECOC 2016; Dusseldorf, Germany, 2016.
20. S. Straullu, M. Befekadu, S. Abrate and R. Gaudino, "Optimization of DSP-based channel
aggregation parameters for front-hauling over PON infrastructure," 2016 IEEE Photonics
Conference (IPC), Waikoloa, HI, 2016.
21. P. Torres-Ferrera, S. Straullu, S. Abrate and R. Gaudino, "Upstream and downstream anal-
ysis of an optical fronthaul system based on DSP-assisted channel aggregation," IEEE/OSA
IEEE/OSA Journal of Optical Communications and Networking (JOCN), vol. 9, no. 12, pp.
1191-1201, Dec. 2017.
22. A. Udalcovs et al., "An Insight into the Total Cost of Ownership of 5G Fronthauling," 2018
20th International Conference on Transparent Optical Networks (ICTON), Bucharest, 2018.
23. G. Arevalo, R. Hincapie, R. Gaudino, "Optimization of multiple PON deployment costs
and comparison between GPON, XGPON, NGPON2 and UDWDM PON", Elsevier Optical
Switching and Networking, vol. 25, pp. 80-90, 2017.
24. ITU-T G.983.4 Recommendation, "A broadband optical access system with increased service
capability using dynamic bandwidth assignment.
Post-OFDM modulations for 5G and beyond
Paolo Banelli, Giulio Colavolpe, Luca Rugini, Alessandro Ugolini
Abstract In this chapter, we present a quick overview of the most popular multicar-
rier modulation techniques that could potentially be adopted in future wireless and
cellular networks. Starting from the widely adopted OFDM, we present strong and
weak aspect of each strategy, describing how it is possible to model all considered
formats by means of a unified signal processing framework. We then propose an
information-theoretical framework for performance evaluation, followed by a few
numerical examples in wireless scenarios. Our analysis shows that the best format
should be selected depending on the channel model and transceiver constraints.
1 Introduction
One of the key technologies that allowed the impressive data rate increase from tens
of kbits/s in 2G systems to the current state of the art of tens of Mbits/s in long
term evolution (LTE) systems has been the evolution from single-carrier modulation
schemes with binary constellations to multicarrier modulations with multilevel con-
stellations [1]. 5G mobile communications aim at delivering gigabit transmission to
mobile users [2]. To achieve such an ambitious goal, new techniques and strategies
must be applied to different network layers and aspects. At the physical layer, an
intense research activity has been dedicated to the study of different modulation for-
mats (see [3] and references therein). Orthogonal frequency division multiplexing
(OFDM) and its natural extension OFDMA are the modulation and multiple access
format adopted by the current LTE standard; OFDM offers a series of attractive
properties and features, that made it widely popular, but it is not exempt of defects
and drawbacks.
Paolo Banelli, Luca Rugini

University of Perugia, Italy, e-mail: paolo.banelli@unipg.it, luca.rugini@unipg.it
Giulio Colavolpe, Alessandro Ugolini
University of Parma, Italy e-mail: giulio.colavolpe@unipr.it, alessandro.ugolini@unipr.it
109
110 Paolo Banelli, Giulio Colavolpe, Luca Rugini, Alessandro Ugolini
Hence, the aim of this chapter is twofold. First, we will provide a short overview
of the most credible competitors of OFDM, detailing for each of them the strong
and weak aspects. Then, we will offer an information-theoretical framework that
allows us to compare the different formats using a common tool, based on a unified
signal processing description. At the end of the chapter we will show some examples
of performance comparison of the proposed waveform techniques using standard
wireless channel models.
2 Modulation Formats
In this section, we briefly introduce OFDM [4, 5] and three alternative schemes,
namely filterbank multicarrier (FBMC) [6, 7], generalized frequency-division mul-
tiplexing (GFDM) [8], and universal filtered multicarrier (UFMC) [9, 10]. For each
scheme, we provide a general description and we underline the strong and weak
points. We refer the reader to [11] for a complete mathematical formulation of the
different modulation formats. Moreover, as demonstrated in [11], all these modula-
tion schemes can be represented by means of a unique discrete-time model, which
allows to easily compare them and to define a common framework for their perfor-
mance evaluation (see [11] for the details).
2.1 OFDM
OFDM is a popular multicarrier technique widely employed for wireless communi-

cations and specifically for 4G cellular systems [4, 5]. A key feature of OFDM is the
simple generation of the transmitted signal by means of inverse fast Fourier trans-
form (IFFT) processing and the simple recovery of the received data by means of fast
Fourier transform (FFT) processing. For each OFDM block, a cyclic prefix (CP) is
inserted at the transmitter and removed at the receiver. Therefore, if the duration of
the multipath channel does not exceed the duration of the cyclic prefix, consecutive
OFDM data blocks do not overlap in the time domain: hence, the OFDM trans-
mission and reception can be performed independently on each block. In addition,
OFDM with CP tranforms a frequency-selective multipath channel into a parallel set
of frequency-flat single-path channels, thereby enabling a simple equalization that
uses only a scalar gain compensation for each subcarrier. The reduced-complexity
equalization of OFDM is one of the main factors of the popularity of OFDM for
multipath channels. Besides, OFDM is suitable to multiple-input multiple-output
(MIMO) systems based on multiple antennas, leading to a separate MIMO channel
for each subcarrier. A variant of OFDM consists in replacing the CP with zeros, and
it is denoted as zero-padding (ZP) OFDM [12].
One of the drawbacks of OFDM is the loss of spectral efficiency and capacity
caused by the CP or the ZP. Furthermore, the spectral sidelobes of the signal transmit-
Post-OFDM modulations for 5G and beyond 111
ted in each subcarrier are non-negligible: therefore OFDM is sensitive to frequency

offsets and Doppler effects, which destroy the frequency orthogonality, leading to
intercarrier interference (ICI). Another weakness of OFDM (and of multicarrier
techniques) is the relevant nonlinear distortion caused by high-power amplification
at the transmitter side: the signal to be transmitted is the IFFT combination of the
signals of many subcarriers, and hence the transmitting amplifier has to deal with a
signal with large peak-to-average power ratio.
2.2 FBMC
FBMC [6, 7] is a multicarrier scheme with some key differences with respect to
OFDM. A specific feature of FBMC is that consecutive data blocks overlap in time:
with respect to OFDM, the duration of the prototype filter applied to each FBMC
data block is increased. A longer prototype filter yields a signal with larger duration
and allows for reduced frequency-domain sidelobes. As a consequence of the low
sidelobes, when frequency offsets or time synchronization errors are present, FBMC
collects a reduced ICI with respect to OFDM. For the same reason, FBMC generates
less adjacent channel interference (ACI) than OFDM: this simplifies the coexistence
with other signals allocated in nearby bands. Hence, FBMC is suitable for multiple-
access and cognitive radio applications.
On the other hand, FBMC requires more advanced data detection techniques, with
respect to OFDM, because of the time-domain overlap of consecutive data blocks:
FBMC intentionally introduces intersymbol interference (ISI) between consecutive
data blocks and ICI, also in the absence of frequency offsets. In other words, the
non-orthogonality of FBMC produces ISI and ICI. By comparison, OFDM uses a
prototype filter with duration equal to the duration of a data block, thereby avoiding
the ISI between consecutive data blocks by separating the waveforms of different
data blocks in the time domain; in addition, OFDM avoids the ICI in the absence
of frequency offsets, because of its carriers orthogonality. The advanced detection
techniques required by FBMC can be interpreted as ISI mitigation/cancellation
techniques, with additional complexity with respect to OFDM, whose detection is
independently performed on each single block.
2.3 GFDM
GFDM [8] is a multicarrier method that combines the advantages of both OFDM
and FBMC techniques. Indeed, like in FBMC, GFDM uses prototype filters with
duration larger than the duration of a data block: this reduces the spectral sidelobes
with respect to OFDM. However, differently from FBMC, multiple data blocks of
GFDM constitute a single superblock that does not interfere with the adjacent ones
because the prototype filters of GFDM are chosen to avoid the time-domain overlap
among consecutive superblocks. Besides, similarly to OFDM, GFDM adopts a CP

to separate the consecutive superblocks and to reduce the equalization complexity in
multipath channels: the CP of GFDM is inserted on a superblock basis, that is, the
same CP is shared by multiple data blocks. This way, compared to OFDM, GFDM
reduces the spectral efficiency loss caused by the CP insertion. In summary, GFDM
is a non-orthogonal data-block technique that aims at reduced sidelobes (like FBMC
schemes) and keeps time-domain orthogonality and per-subcarrier equalization on
a data-superblock basis (like CP-based schemes).
The main drawback of GFDM is its increased complexity with respect to OFDM,
since the signal processing has to be performed on a data superblock basis, rather
than on a data block basis like in OFDM. With respect to FBMC, if we assume the
same duration of the prototype filters, the complexity of GFDM has the same order
of magnitude (approximately).
2.4 UFMC
UFMC [9, 10] is a multiuser multicarrier technique where each user employs a
subband with consecutive subcarriers. The main goal of UFMC is to reduce the ICI
with respect to OFDM and its multiple access counterpart OFDMA, which simply
distributes carriers among users. The key idea of UFMC is to adopt prototype filters
that reduce the sidelobes on a subband basis: differently, FBMC aims at the same
purpose on a subcarrier basis. This choice in UFMC produce prototype filters whose
duration is significantly reduced with respect to that of FBMC. Typically, the UFMC
filter duration is similar to the CP duration of OFDM, that is, by far lower than the
data block duration; instead, the FBMC filter duration is an integer multiple of the
data block duration.
Like in OFDM, the UFMC data blocks do not overlap in time at the transmitter:
however, in multipath channels, since UFMC does not employ a CP, a limited amount
of ISI between consecutive data blocks is present. With respect to OFDM, another
weakness of UFMC is the increased equalization complexity. Anyway, since the
UFMC signal detection is performed on a data block basis, the UFMC equalization
complexity is generally lower than for FBMC. Note that the UFMC subbands have
adjacent subcarriers: on the other hand, in frequency-selective channels, frequency
diversity considerations would suggest a frequency interleaved allocation where each
user exploits maximally separated subcarriers.
3 Equalization and Detection Strategies
In this section, we discuss some equalization and detection strategies that are common
to all the previously described modulation techniques. Based on the unified discrete-
time model [11], we can express the received samples in a compact vector form,
as a function of a system matrix, which takes into account a possible transmitter

processing and the propagation channel, the transmitted data on all subcarriers, and
the thermal noise (refer to [11] for the details). We remark again that, depending on
the modulation format, different kinds of interference can arise, and these can all be
modeled by properly designing the system matrix in the unified model.
Different equalization options can be envisaged to cope with the arising interfer-
ence. We briefly present some of them in the list below.
• Least square (LS) linear equalizers can be designed to remove only the ICI, or
both ICI and ISI. In the former case, ISI is treated as additional noise, and the
ISI-plus-noise term is often amplified. The latter equalizer design tries to jointly
remove both ICI and ISI, resulting in better performance at the cost of an increased
complexity.
• Alternative to LS, linear minimum mean-square error (MMSE) equalizers are
less sensitive to the noise enhancement phenomenon, and, like LS equalizers, can
be designed to remove ICI only, or ICI and ISI jointly.
• Many other linear equalization techniques can be designed, which often involve
variants or modifications of the previous approaches [13].
• Popular nonlinear equalization approaches involve the use of interference cancel-
lation techniques [14, 15, 16, 17], usually performing the following four steps:
a) preliminary estimation of the data by a linear equalizer; b) reconstruction of
the interference based on the preliminary estimates; c) hard (or soft) cancellation
of the interference, performed by subtracting the reconstructed interference from
the received signal; d) final equalization by a second linear equalizer. Interference
cancellation can be done either serially, or in parallel, or in groups. In serial can-
cellation, a single interference subcarrier is estimated and subtracted, and then
the same process is repeated for the other data subcarriers, in a decision-feedback
way. In parallel cancellation, all interference subcarriers are preliminarly esti-
mated and subtracted jointly, and then all the data subcarriers are detected using a
second equalizer. In group cancellation, subcarriers are divided in groups: the in-
terference among different subcarriers of the same group is cancelled in a parallel
way, while the interference among different groups of subcarriers is cancelled in
a serial way. In all cases, turbo approaches are also possible, to iteratively refine
the equalization of the data already equalized [18, 19].
• An alternative nonlinear equalization method is based on maximum likelihood
(ML) techniques [20, 13]. The key idea of ML equalization is to exploit the finite
alphabet of the data symbols. From a performance viewpoint, ML equalization
is a promising approach: for equiprobable data symbols, the ML detector mini-
mizes the probability of error detection and hence yields the best performance.
However, the main weakness of this approach is its computational complexity,
which depends on the chosen transmission scheme and on several system param-
eters, like the number of subcarriers, the number of interfering symbols, and the
size of the adopted constellations. The complexity of the ML detector can be
reduced by exploiting the special structure of the transmitted signal (which might
be designed to avoid the presence of ISI and/or ICI, depending on the adopted
transmission scheme), or by reducing the search space by excluding impossible or
unlikely combinations of symbols [21]. However, the complexity of ML equaliza-

tion often makes this approach unfeasible in multicarrier wireless communication
systems.
4 Channel Capacity and Spectral Efficiency
Rather than focusing on complexity issues, we are more interested in comparing the
different transceiver architectures from a performance viewpoint. Several alternative
figures of merit can be used to evaluate and compare different modulation formats.
Among these, we can think of computing the bit error rate (BER) performance in
different scenarios or channel conditions. The downside of this perspective resides
in being dependent on the specific adopted code, which might be unsuitable for all
application scenarios, or might require a particular design or optimization. To avoid
this problem, and to offer a wider perspective on the different modulation formats,
we intend to provide the reader with the instruments to compare different transceiver
architectures from an information-theoretic point of view. Our aim is to take into
account that different non-orthogonal waveforms may use the time-frequency re-
source in different ways, may introduce (and may be able to tolerate) a different
amount of interference, and may (or may not) enable suboptimal low-complexity
receivers. As a figure of merit we will use the achievable spectral efficiency (ASE),
with the constraint of arbitrarily small BER. The ASE is computed by dividing the
channel capacity, or an achievable lower bound of it, by the employed symbol time
and frequency spacing, which are the time-frequency resource of every waveform.
As far as the computation of the channel capacity is concerned, we are mainly
interested in the achievable performance when using suboptimal low-complexity
detectors. Therefore, we consider simple receivers based on linear processing fol-
lowed by symbol-by-symbol detection, using the framework described in [22, 23].
This framework allows to compute a lower bound on the channel capacity (and thus
on the ASE) by substituting the actual channel with an arbitrary auxiliary channel
that has the same input and output alphabets of the original channel. This approach
is called mismatched detection theory [22, 23]. An auxiliary channel that approx-
imates the true channel with increased accuracy (with respect to another auxiliary
channel) produces a lower bound that is closer (with respect to another auxiliary
channel) to the true capacity. If the considered suboptimal receiver is optimal for the
adopted auxiliary channel, the obtained lower bound is achievable by that detector,
according to the mismatched detection theory [23]. Therefore, when that auxiliary-
channel-optimal receiver is employed, we say that the computed lower bound is the
ASE of the considered channel with the considered waveforms and receivers. When
Gaussian inputs are considered, closed-form achievable lower bounds of the ASE
can be provided. The same mismatched detection framework can also be used when
finite constellations are employed, but in this case no closed-form expressions for
the lower bounds exist: these lower bound expressions have to be computed numer-
ically, by feeding the auxiliary-channel-optimal detector with the output of the true
channel [23].
Depending on different receiver architectures and different assumptions on the
channel model, we can resort to different auxiliary channels. Each auxiliary channel
will result in a different lower bound on the channel capacity. Herein we neglect
interference-cancellation-based receivers, because it is not possible to find an auxil-
iary channel for which these receivers are optimal: for these receivers the principle of
mismatched detection cannot be adopted [22]. We focus on the following auxiliary
channel models.
• Vector input, vector output, before equalization. The use of multicarrier modula-
tion formats allows us to exploit channel capacity results related to multiantenna
systems. We can model the received samples as a classical MIMO channel, char-
acterized by a channel matrix which multiplies the transmitted data, followed by
the addition of the noise. In the absence of an equalizer, ISI cannot be removed,
so it must be taken into account by properly modifying the covariance matrix of
the noise term. Assuming Gaussian input symbols, the mutual information for
this auxiliary channel can be computed in closed form [11].
• Vector input, vector output, after equalization. The use of a linear equalizer
is expected to enhance the contribution of the useful signal component and,
possibly, to reduce the contribution of the interference, at the expense of some
noise amplification. Basically, a linear equalizer (LS or MMSE, for example)
consists in multiplying the observed vector by a properly designed matrix. Also
in this case, the use of Gaussian symbols allows to compute in closed form the
mutual information for this auxiliary channel model [11].
• Scalar input, scalar output, after equalization. If, instead of considering all sub-
carriers at the same time, we focus on the data symbol transmitted on a single
subcarrier, we can obtain a scalar auxiliary channel model. In this case, all symbols
transmitted on the other subcarriers are regarded as interference, and considered
as additional noise. Also in this case, the use of a linear equalizer can help to
cope with the interference, but the fact that the receiver operates on a symbol-by-
symbol basis, neglecting the possible correlations among the various elements,
tells us that, with this scheme, we should expect worse performance compared
to the equalized vector model. Assuming Gaussian distributed symbols, a closed
form for the mutual information is available also for this case [11]. On the other
hand, when input symbols belong to a finite constellation, as is the case with
practical transmission schemes, no closed form exists, but we can compute the
mutual information by numerical simulations [24].
The expressions for the mutual information that we can compute through the pre-
viously described techniques are based on a single channel realization. To take into
account the effects of a block fading channel model (as usual in wireless communi-
cations scenarios), we can define the ergodic mutual information by computing the
statistical average with respect to the channel realization of the mutual information
expression. This problem can be solved in a semi-analytical way by performing a
numerical average of the mutual information over many channel values.
We can define the ASE as the ratio between the obtained ergodic mutual informa-
tion (which, we recall, is an achievable lower bound to the channel capacity) and the
product between the frequency spacing between subcarriers and the symbol interval
of each subcarrier. Using the ASE as a performance measure allows a fair compari-
son between orthogonal and non-orthogonal waveforms, since it represents exactly
the amount of information that can be transmitted in the units of time and frequency.
Typically, non-orthogonal formats will reduce the achievable mutual information,
due to the increased interference. However, they also reduce the time and/or fre-
quency occupation, so their overall ASE can be higher than that achievable with
orthogonal formats. In general, it is necessary to optimize the values of the time and
frequency spacings to obtain the best performance [25]. When the channel is slowly
varying or constant over a block, it is likely that there will be blocks for which it
is impossible to achieve an arbitrarily low error probability. In these circumstances,
the channel is defined to be in outage, and a more significant measure is the outage
capacity [26]. Due to the space limitations, we will not consider this aspect further.
5 Numerical Results
In this section, we compare the different signal waveforms in terms of ASE. The
results reported in this chapter are far from being exhaustive of all the possible
use cases and scenarios that can arise in 5G systems. We refer the reader to [11]
for a more comprehensive set of results. Here, we limit our analysis to two cases
representative of a single-antenna downlink system, which are sufficient to give
meaningful insights on the potential of each signal waveform. Regarding the channel
models, we adopt the classical and general multipath fading models, like those with
sparse power-delay profiles, selected according to LTE channel models [27], and
those with exponentially decaying profiles, widely used for WLAN applications [28].
The ASE of the different waveform schemes massively depends on a large set of
parameters, such as the possible CP length and the number of subcarriers, as well as
the parameters characterizing the adopted shaping pulses, which change depending
on the selected scheme. A full parameters optimization is outside the scope of this
chapter; the interested reader can find in [11] the details on the selected parameters
and the rationale of these choices. For the results in this section, we just select the
best parameters configurations among those considered in [11].
We will compare the ASE of the different signal waveforms by semianalytically
computing the lower bounds derived in the previous section. We will consider only the
vector and scalar bounds computed after MMSE equalization, since we have verified
that the absence of the equalizer does not change the performance significantly [11].
For a fair comparison, the scalar bound is summed over all the symbols constituting
the vector used for the vector model.
Figs. 1 and 2 compare the ASE of the considered techniques on the sparse
Extended Pedestrian A (EPA) channel [27], and on the WLAN A exponential
model [28], as a function of the signal-to-noise ratio (SNR). Even from this limited
set of results, we can clearly evince that the best technique depends on the channel
model. On the EPA channel, ZP-OFDM achieves the best performance, followed by
OFDM and UFMC. The loss of OFDM (with respect to ZP-OFDM) is due to the
additional energy spent for transmitting the CP: this loss is reduced, since the EPA
channel has a small delay spread, hence a short CP is sufficient. Also GFDM and
UFMC give good ASE results: in case of UFMC, the ASE is quite large also for
the scalar bound. On the other hand, FBMC and the scalar bound of GFDM give
inferior ASE results, due to the presence of interference. Therefore, for the EPA
channel, orthogonal and non-overlapping multicarrier techniques like OFDM and
UFMC are preferable with respect to overlapping methods like GFDM and FBMC.
Fig. 2 exhibits the ASE comparison in the WLAN channel A. In this scenario, the
best ASE performance is achieved by UFMC. However, the ASE performance of
the orthogonal techniques (OFDM and ZP-OFDM) is still acceptable. On the other
hand, the ASE performance of the overlapping techniques (FBMC and GFDM) is
quite good at low SNR: at high SNR, the FBMC or GFDM receiver should target
the vector bound.
In conclusion, there is no single winner in all the cases, at least from the ASE
viewpoint. The choice among different techniques can be driven by the specific
channel scenario. In addition, the complexity of the receiver should be taken into
account, to grant such a capacity promise. Orthogonal methods (OFDM and ZP-
OFDM) produce good results in all scenarios, except when the length of the CP or
ZP is excessive. Overlapping techniques (FBMC and GFDM) promise larger ASE
if the receiver can achieve the vector bound: however, in many scenarios, the scalar
bound for overlapping techniques is reduced with respect to OFDM. UFMC is able
to outperform OFDM in some scenarios. Summarizing, no specific technique is able
to largely outperform the competitors in all scenarios.
6 Concluding Remarks
In this chapter, we have provided a brief overview of the most popular multicarrier
waveforms proposed for 5G cellular systems and beyond. Basing our analyses on
a unified signal processing framework, able to encompass all modulation formats,
we proposed to compare and evaluate the performance of the different techniques
from an information-theoretical point of view, based on the computation of the ASE,
which allows to obtain a fair comparison of the alternative formats without being
constrained to adopt a fixed coding scheme. Our results show, unsurprisingly, that the
final choice of a specific signal waveform should depend on the channel conditions,
as well as on the system requirements and available computational complexity.
Fig. 1 ASE versus SNR on the EPA channel.
Fig. 2 ASE versus SNR on the WLAN A channel.

References
1. ETSI, “LTE; Evolved Universal Terrestrial Radio Access (E-UTRA); Base Station (BS) radio
transmission and reception (3GPP TS 36.104 version 11.6.0 Release 11),” Oct. 2013.
2. J. G. Andrews et al., “What will 5G be?” IEEE J. Select. Areas Commun., vol. 32, no. 6, pp.
1065–1082, Jun. 2014.
3. P. Banelli et al., “Modulation formats and waveforms for 5G networks: Who will be the heir
of OFDM?” IEEE Signal Processing Mag., vol. 31, no. 6, pp. 80–93, Nov. 2014.
4. T. Hwang et al., “OFDM and its wireless applications: a survey,” IEEE Trans. Veh. Technol.,
vol. 58, no. 4, pp. 1673–1694, May 2009.
5. Z. Wang and G. B. Giannakis, “Wireless multicarrier communications: Where Fourier meets
Shannon,” IEEE Signal Processing Mag., vol. 17, no. 3, pp. 29–48, May 2000.
6. M. Bellanger, “FBMC physical layer: A primer,” PHYsical layer for DYnamic AccesS and
cognitive radio (PHYDYAS), FP7-ICT Future Networks, Tech. Rep., June 2010.
7. B. Farhang-Boroujeny, “OFDM versus filter bank multicarrier,” IEEE Signal Processing Mag.,
vol. 28, no. 3, pp. 92–112, May 2011.
8. N. Michailow et al., “Generalized frequency division multiplexing for 5th generation cellular
networks,” IEEE Trans. Commun., vol. 62, no. 9, pp. 3045–3061, 2014.
9. F. Schaich and T. Wild, “Waveform contenders for 5G – OFDM vs. FBMC vs. UFMC,” in
Proc. IEEE Intern. Symp. on Commun., Control and Signal Processing, 2014, pp. 457–460.
10. V. Vakilian et al., “Universal-filtered multi-carrier technique for wireless systems beyond LTE,”
in Proc. IEEE GLOBECOM Workshops, 2013, pp. 223–228.
11. P. Banelli, G. Colavolpe, L. Rugini, and A. Ugolini, Information-Theoretic Perspectives on 5G
Systems and Beyond. Cambridge University Press, in press, ch. Waveform Design, eds.: I.
Maric, O. Simeone, and S. Shamai (Shitz), to appear.
12. B. Muquet et al., “Cyclic prefixing or zero padding for wireless multicarrier transmissions?”
IEEE Trans. Commun., vol. 50, no. 12, pp. 2136–2148, Dec. 2002.
13. L. G. Baltar et al., “MLSE and MMSE subchannel equalization for filter bank based multi-
carrier systems: Coded and uncoded results,” in Proc. 18th European Signal Processing Conf.
(EUSIPCO), Aug 2010, pp. 2186–2190.
14. A. Stamoulis et al., “Intercarrier interference in MIMO OFDM,” IEEE Trans. Signal Process-
ing, vol. 50, no. 10, pp. 2451–2464, Oct. 2002.
15. N. Benvenuto et al., “Single carrier modulation with nonlinear frequency domain equalization:
An idea whose time has come again,” Proc. of the IEEE, vol. 98, no. 1, pp. 69–96, Jan. 2010.
16. X. Cai and G. B. Giannakis, “Bounding performance and suppressing intercarrier interference
in wireless mobile OFDM,” IEEE Trans. Commun., vol. 51, no. 12, pp. 2047–2056, Dec. 2003.
17. L. Rugini et al., “Low-complexity banded equalizers for OFDM systems in Doppler spread
channels,” EURASIP Journal on Applied Signal Processing, vol. 2006, pp. 1–13, Aug. 2006.
18. K. Fang et al., “Low-complexity block turbo equalization for OFDM systems in time-varying
channels,” IEEE Trans. Signal Processing, vol. 56, no. 11, pp. 5555–5566, Nov. 2008.
19. P. Schniter, “Low-complexity equalization of OFDM in doubly selective channels,” IEEE
Trans. Signal Processing, vol. 52, no. 4, pp. 1002–1011, Apr. 2004.
20. R. Zakaria and D. Le Ruyet, “On maximum likelihood MIMO detection in QAM-FBMC
systems,” in Proc. IEEE Intern. Symp. on Personal, Indoor, and Mobile Radio Commun.,
2010, pp. 183–187.
21. R. Zakaria et al., “On ISI cancellation in MIMO-ML detection using FBMC/QAM modula-
tion,” in Proc. IEEE Int. Symp. on Wireless Commun. Systems (ISWCS), 2012, pp. 949–953.
22. N. Merhav et al., “On information rates for mismatched decoders,” IEEE Trans. Inform. Theory,
vol. 40, no. 6, pp. 1953–1967, Nov. 1994.
23. D. M. Arnold et al., “Simulation-based computation of information rates for channels with
memory,” IEEE Trans. Inform. Theory, vol. 52, no. 8, pp. 3498–3508, Aug. 2006.
24. R. E. Blahut, Principles and Practice of Information Theory. Addison-Wesley Longman
Publishing Co., Inc., 1987.
25. A. Barbieri et al., “Time-frequency packing for linear modulations: Spectral efficiency and
practical detection schemes,” IEEE Trans. Commun., vol. 57, pp. 2951–2959, Oct. 2009.
26. E. Biglieri et al., “Fading channels: Information-theoretic and communications aspects,” IEEE
Trans. Inform. Theory, vol. 44, no. 6, pp. 2619–2692, 1998.
27. J. A. del Peral-Rosado et al., “Evaluation of the LTE positioning capabilities under typical
multipath channels,” in Proc. IEEE 6th Advanced Satellite Multimedia Systems Conf. (ASMS)
and 12th Signal Processing for Space Commun. Workshop (SPSC), 2012, pp. 139–146.
28. A. Doufexi et al., “A comparison of the HIPERLAN/2 and IEEE 802.11a wireless LAN
standards,” IEEE Commun. Mag., vol. 40, no. 5, pp. 172–180, 2002.
Softwarization and Virtualization
Giacomo Verticale, Antonio Capone
Management and deployment of 5G technology is greatly simplified by the adoption

of a set of complementary technologies enabling flexible usage of deployed hardware
and fast provisioning of new functions and services, namely: Software Defined
Networking (SDN) and Network Function Virtualization (NFV). These, in turn, are
the building blocks for more sophisticated services such as Cloud-RAN, Network
Slicing, and Multiaccess Edge Computing (MEC).
1 Software Defined Networking (SDN)
SDN is an umbrella term to indicate a number of protocols and interfaces allowing

network programmability. This is achieved by decoupling the control and user plane.
In the SDN concept, the user plane is greatly simplified and consists in stateless,
distributed forwarding tables performing packet switching at very high speed. The
tables are populated by a centralized control plane maintaining end-to-end path
information for each service and providing support for advanced functions such as
mobility management, policy and subscription control [14].
Figure 1 shows the basic SDN model according to Open Networking Foundation
(ONF) [13]. A service consumer exchanges both data and management-control
operations with some SDN provider. Service data is ultimately forwarded by some
set of resources that are owned by the SDN provider. The service consumer controls
them through the SDN controller by invoking actions on a set of virtual resources
that it perceives to be its own.
121
122 Giacomo Verticale, Antonio Capone
Service invocation and control
Service
consumer R R
R
SDN Controller
R Resource
R
Service data group
R
Fig. 1 Basic SDN model according to ONF [13]
2 Network Function Virtualization (NFV)
The NFV architecture defines how software functions can be executed in virtual
machines (VMs) and consolidated to share common physical resources in terms of
compute, storage, and networking. Multiple functions can be instantiated within the
same VM using containers [12]. VMs can be dynamically instantiated to cope with
changing network demand in terms of traffic and in terms of offered features and
services. Examples of network functions that can be virtualized include:
• Evolved Packet Core (EPC) functions, including the Mobility Management Entity
(MME), the Serving Gateway (S-GW), and the Packet Data Network Gateway
(P-GW).
• Baseband Processing Unit (BBU) functions
• Switching functions
• Traffic load balancing
With NFV, services are described as sequences of network functions that process
end-to-end flows. Figure 2 shows an example forwarding graph for a mobile Internet
service provider. Data flows from the evolved eNodeB to the service gateway and to
the IP backbone. Mobility management, authentication and other control protocols
flow through different network functions. Unlike cellular networks, where a particular
feature is activated network-wide, 5G enables the operator to activate a feature on a
per-service basis.
NFV and SDN do not require each other, but are related in many ways. SDN
provides a natural solution to route packets between the Virtual Network Functions
(VNFs) that characterize each service. Additionally, it enables the virtualization of
routing functions with a low overhead. Finally, it simplifies the rerouting of traffic
flows after a particular VNF is moved from one physical node to another or, similarly,
when an additional instance of a VNF is elastically deployed in a new node to cope
with increasing traffic demands.
Softwarization and Virtualization 123
subscriber data
network access protocols
application data
VNF
VNF VNF Load Balancer
eNodeB HSS
VNF
VNF
VNF MME VNF VNF Firewall
eNodeB S-GW P-GW Application
Mobile Server
terminal Virtualization Layer
Internet
Fig. 2 Example of NFV forwarding graph [8]
The main challenge for NFV solutions is complexity. In order to be cost-effective,

many VNFs must be consolidated in a single physical node. As a result, the physical
server must be able to manage multiple traffic streams with possibly overlapping
addresses while keeping overhead to a minimum.
3 Cloud-RAN
The scenario in which baseband processing is implemented in virtual Baseband

Units (BBUs) is known as Cloud-RAN [9]. The virtualized infrastructure manager
deploys a pool of virtual BBUs. The cell site simplifies to antennas, Remote Radio
Units (RRUs), and switching functions. The switching functions interconnect the
virtual BBUs to the RRUs via high-speed optical links to meet latency requirements.
According to traffic demand, the VNF Manager allocates BBUs to active cell sites
and programs an overlay virtual network to switch traffic flows from the cell site’s
RRU to the VMs hosting the allocated BBU.
According to [8] Cloud-RAN yields savings over a non virtualized architecture
both in terms of CAPEX and OPEX. CAPEX savings stem from the fact that a
single virtual BBU can serve traffic from multiple cell sites. Thus, the total number
of required BBUs depends on the maximum traffic of the network rather than the
maximum traffic of each individual site. OPEX savings stem from the fact that the
energy consumption of the whole system depends only on the average number of
active BBUs, which depends on the average network traffic rather than on the average
number of active cell sites.
Additionally, Cloud-RAN helps the deployment of ultra densified networks. In
these deployments, the mobile terminal connects to the network through a cluster
of closest cells, which must cooperate to minimize inter-site interference through

inter-site scheduling. The consolidation of BBUs in a single physical site makes such
cooperation easier and faster.
4 Network Slicing
5G is designed to be a multi-service network supporting multiple verticals with a

diverse set of performance requirements. The key to realize this vision is slicing
the physical network into multiple isolated logical networks on a per-service basis.
Network slicing is the technical mean for allowing the coexistence of different
verticals over the same infrastructure.
Services
5G Edge 5G Core IP Service Network
Fig. 3 Example of 5G network supporting multiple network slices
According to the Next Generation Mobile Networks (NGMN) Alliance, a network

slice is a set of network functions, and the resources to run these functions, forming
a logical network that meets the requirements of a given service[11]. Figure 3
exemplifies the NGMN definition of network slice.
Network Slicing opens the way to a model in which multiple stakeholders coop-
erate to provide the final service. The 5G-PPP identifies three such roles [7]:
• The tenant offers an end-to-end service to the final users and expresses the
requirements of the network slice.
• The Mobile Service Provider (MSP) brokers the resources, either internally or
from the external Infrastructure Providers, manages the network slices, and is
responsible of ensuring that each network slice meets the tenant’s requirements.
It may operate as a Network-as-a-Service (NaaS) or as a Platform-as-a-Service

(PaaS) provider.
• the Infrastructure Provider provides the resources, which can be either physical
resources or virtualized resources in a shared network or data center.
Service Layer
Control Layer Management and

Orchestration
Data Layer Layer
Physical Virtualized
Infrastructure Infrastructure
Fig. 4 Layers of the 5G architecture [7]
5G-PPP architectural vision is depicted in Figure 4 and reflects the existence of

different business roles and business relationships. It comprises different layers:
• The Service Layer comprises the functions for the creation of value, namely
the logic and data of the various applications using the 5G infrastructure. It also
comprises the tools for analyzing big data and making predictive decisions in
real-time.
• The Management and Orchestration Layer (MANO) includes the functions
for the management of the life cycle of Virtual Network Functions (VNFs) and
allocation of resources. It transforms customer-oriented service descriptions into
resource-oriented service description.
• The Control Layer transforms service requests into configurations and com-
mands to Physical and Virtual Network Functions. It is responsible of routing
decisions and of ensuring that the service meets the performance requirements.
• The Data Layer comprises the Physical and Virtual Network Functions responsi-
ble for value enablement. It includes common networking functions such as user
authentication, resource accounting, or network security and dedicated, value-
added or mission specific services such as video delivery, state prediction, facility
monitoring, or machine control.
• The Physical Infrastructure includes assets such as compute, network, and
storage, which are distributed in the back-end data centers, in the core network
infrastructure, and radio access network.
• The Virtualized Infrastructure consists of compute, network, and storage re-
sources reserved in a shared network or data center.
This architecture yields several options for the design of network slices. Figure 5
shows some of these options: standalone slices with own hardware, slices with own
spectrum, which share only RF and antenna equipment, slices with shared resources
Resource Resource
Slice with unaware unaware
Slice with shared slice slice
Standalone own resources
slice spectrum
Common PDCP / RLC
Common baseband processing
Common RF and antenna
Fig. 5 Options for slice multiplexing
up to the baseband processing, and resource-unaware slices, which have no control

over the assigned resources.
5 Multiaccess Edge Computing
The idea of a computing platform located at the network edge is not specific to the
world of cellular networks. This is the reason why ETSI choose the name Multiaccess
Edge Computing (MEC) for its working group in charge of standardizing such
platforms. A MEC platform follows the trend towards cloud-based architecture, but
exploits the advantage of being located in close proximity to the end users. In the
case of 5G, this means in the RAN. MEC architectures take advantage of the existing
NFV infrastructure but are further characterized by low latency, proximity, location
awareness, high bandwidth, and real-time insight into radio network information.
Exploiting real-time location and radio conditions can create a context for a new
ecosystem of use cases with an improved user experience.
In order to support Cloud-RAN (see Section 3), operators will deploy small
datacenters at the edge. These IT infrastructures will make it possible to deploy
applications and VNFs at the edge without investing additional resources. As a
consequence, MEC will become cheaper and more appealing. Those operators that
also have fixed subscribers will find natural to concentrate mobile and network
operations at the same Points-of-Presence (PoPs), making these locations candidates
for providing edge computing for all customers. The MEC platform at each data
center will allow third party applications to activate traffic offloading at the edge
while also enabling access to other information provided by the operator.
6 Software Platforms
There are several efforts for implementing a NFV-capable platforms, often based on
the ETSI MANO specifications [10]. A growing number of such frameworks are
being released with an open source licence, either because they are the result of a
publicly-funded research project, or they were commercial projects later donated to
the community in an attempt to stimulate the adoption. We briefly discuss the most
relevant projects in this class.
• OSM – Opensource MANO is an ETSI-hosted project aiming at being the reference
implementation of the ETSI MANO specifications.[5]
• ONAP – Open Network Automation Platform is a Linux Foundation project that
is rapidly growing and has strong industrial support. The initial release includes
two blueprints: Voice over LTE, and Residential vCPE.[2]
• Open Baton is an implementation of ETSI MANO provided by Fraunhoefer
FOKUS.[3]
• SONATA is the output of an EU-funded research project and provides an SDK for
developing VNFs.[6]
• OPNFV is another Linux Foundation project. OPNFV aims at building an ETIS
MANO platform by integrating components from upstream projects [4]
• M-CORD is an ONF hosted project and provides both virtualization of RAN
functions and a virtualized mobile core (vEPC) to enable mobile edge applications
and innovative services using a micro-services architecture.[1]
7 Conclusion
Communication networks are quickly adopting the cloud model that revolutionized
the IT world, providing both flexibility and cost reduction through consolidation of
the infrastructure. 5G is adopting network virtualization as a key technology both in
the RAN and in the core network.
Operators will deploy small data centers in the RAN, at the edge of the network.
These data centers will enable virtualization of the baseband processing, resulting
in significant savings. They will also support MEC, which will enable new services
for the end user characterized by low latency and location awareness.
Network slicing, which virtualizes both the radio access and core networks, will
enable per-service performance levels and isolation. It will support a model in which
the mobile service operator operates as a resource broker, which pools resources
from different infrastructure operators and provides network slices. Each slices is
then operated by a tenant, which offers the service to the end-user.
In addition to 3GPP, several organizations are standardizing network softwariza-
tion and virtualization protocols and interfaces, most notably ETSI and ONF. In
parallel to standardization activities, several projects are releasing components un-
der an open-source licence. These components are accelerating the adoption of
newer technologies and are becoming reference implementations, thus improving

interoperability.
There are many issues still open and which require more operational experience
to be fully understood. Some of these issue are: how to distribute network functions
over different execution platforms by different vendors; how to efficiently and reliably
translate service requirements into resource requirements; how to provide a trusted
execution environment for third party services over shared resources; how to manage
the operational complexity of a large network with multiple stakeholders.[15]
References
1. Mobile Central Office Re-architected as a Datacenter (M-CORD). URL

www.opennetworking.org/m-cord
2. ONAP – Open Network Automation Platform. URL www.onap.org
3. Open Baton. URL openbaton.github.io
4. Open Platform for NFV (OPNFV). URL www.opnfv.org
5. OSM – Open Source MANO. URL osm.etsi.org
6. SONATA NFV: Agile Service Development and Orchestration in 5G Virtualized Networks.
URL www.sonata-nfv.eu
7. 5G PPP Architecture Working Group: View on 5G architecture (version 2.0) (2017)
8. Abdelwahab, S., Hamdaoui, B., Guizani, M., Znati, T.: Network Function Virtualization in 5G.
IEEE Communications Magazine 54(4), 84–91 (2016). DOI 10.1109/MCOM.2016.7452271
9. Checko, A., Christiansen, H.L., Yan, Y., Scolari, L., Kardaras, G., Berger, M.S., Dittmann,
L.: Cloud ran for mobile networks—a technology overview. IEEE Communications Surveys
Tutorials 17(1), 405–426 (2015). DOI 10.1109/COMST.2014.2355255
10. ETSI: GS NFV-MAN 001 V1.1.1 Network Functions Virtualisation (NFV); Management and
Orchestration (2014)
11. Foukas, X., Patounas, G., Elmokashfi, A., Marina, M.K.: Network Slicing in 5G: Sur-
vey and Challenges. IEEE Communications Magazine 55(5), 94–100 (2017). DOI
10.1109/MCOM.2017.1600951
12. Mijumbi, R., Serrat, J., Gorricho, J., Bouten, N., Turck, F.D., Boutaba, R.: Network func-
tion virtualization: State-of-the-art and research challenges. IEEE Communications Surveys
Tutorials 18(1), 236–262 (2016)
13. Open Networking Foundation (ONF): TR-521 SDN Architecture issue 1.1 (2016). URL
www.opennetworking.org
14. Shafi, M., Molisch, A.F., Smith, P.J., Haustein, T., Zhu, P., De Silva, P., Tufvesson, F., Ben-
jebbour, A., Wunder, G.: 5G: A tutorial overview of standards, trials, challenges, deployment,
and practice. IEEE Journal on Selected Areas in Communications 35(6), 1201–1221 (2017).
DOI 10.1109/JSAC.2017.2692307
15. Yousaf, F.Z., Bredel, M., Schaller, S., Schneider, F.: NFV and SDN-Key technology enablers
for 5G networks. IEEE Journal on Selected Areas in Communications 35(11), 2468–2478
(2017). DOI 10.1109/JSAC.2017.2760418
Efficiency/Off Grid Operation
Luca Chiaraviglio, Michela Meo
Abstract In order to be deployed in scenarios where the power grid is not available
and/or not reliable, the 5G technology has to face several challenges, which are
related to its costs and its energy-efficiency. These issues particularly emerge in
rural and low-income areas, where the operators are not keen to deploy the same
5G infrastructure originally designed to serve urban zones, due to low Return On
Investment (ROI) rates, as well as in emerging countries in which the power grid is
not reliable. In this chapter, we face the efficiency and the off grid operation of a
5G network, by: i) defining a 5G architecture based on Unmanned Aerial Vehicles
(UAV) and Large Cells (LCs), ii) analyzing the CAPital EXpenditures and OPex
EXpenditures of the aforementioned architecture, iii) evaluating different strategies
for the reduction of the costs during the design phase, iv) introducing the energy-
efficient management of an UAV-based network, v) considering the introduction of
renewable energy sources. Results, obtained over several case studies, demonstrate
that an efficient 5G off-grid architecture can be deployed, with a positive impact to
the connectivity of the users living in rural and low-income areas; the introduction of
renewable energy sources allows also the off-grid operation in emergency situations.
1 Introduction
One of the peculiar aspects of 5G is its possibility to improve the cost/energy

efficiency compared to pre-5G solutions, as well as the possibility of working in
Luca Chiaraviglio
Department of Electronic Engineering, University of Rome Tor Vergata, Rome, Italy, and
Consorzio Nazionale Interuniversitario per le Telecomunicazioni, Italy e-mail: luca.
chiaraviglio@uniroma2.it
Michela Meo
Department of Electronics and Telecommunications, Politecnico di Torino, Turin, Italy, e-mail:
michela.meo@polito.it
129
130 Luca Chiaraviglio, Michela Meo
Table 1 Comparison of a classical 5G Urban Scenario with Rural and Low-income Ones [1]
5G Urban Scenario 5G Rural Scenario 5G Low-income Scenario
Basic Connectivity, Emergency Ser-
HD Video, HD Streaming, Tactile In- HD Video, Emergency Service, e-
Service Type vice, Delay Tolerant, e-Health, e-
ternet, IoT Health, e-Learning
Learning
Network Con- Maximize Bandwidth, Minimize De-
Coverage, Guaranteed Bandwidth Coverage
straints lay, Coverage
Unreliable Power Grid and/or Renew-
Energy Sources Power Grid Power Grid, Renewable Sources
able Sources
Monthly User
Pay per bandwidth Same as standard urban users Low
Subscription Fee
Business Model Return on Investment Subsidized by the government Subsidized by the government
Required Net-
High High High
work Flexibility
Pedestrian, Vehicular, High Speed Ve-
User Mobility Pedestrian, Vehicular Pedestrian, Low Speed Vehicular
hicular
scenarios where the connection to the power grid is not available or not reliable.
These two requirements emerge in particular in rural and low-income areas, where
the pre-5G networks are currently being not (or not sufficiently) deployed, due to
relatively low Return On Investment (ROI) rates for the operators and in emerging
countries in which the power grid is not reliable. Tab. 1 reports a comparison among
a classical 5G urban scenario, a 5G rural scenario, and a 5G low-income scenario.
Focusing on the service type, people living in rural zones are willing to receive a
service comparable to the one available in urban areas; this includes services like
High Definition (HD) video and good connectivity. In addition, specific services
for the rural areas, including emergency services, e-Health and e-Learning, may be
required. On the other hand, people living in low-income zones are subject to basic
connectivity service requests (rather that HD ones), due to the fact that the adopted
devices in these zones mainly include smartphones rather than (expensive) smart
TV and/or personal computers/laptops. On the other hand, an emergency service is
very important for such zones, which may be coupled with e-Health and e-Learning
activities. In addition, delay tolerant services may be required in the low-income
scenario. Given the requirements in terms of service type, the 5G network constraints
in urban areas include the maximization of the bandwidth and the coverage, which
are coupled with the minimization of delay. In rural areas it is instead very important
to guarantee coverage and a given amount of bandwidth. Moreover, the coverage is
also a stringent constraint in low-income zones. Focusing then on the energy sources,
the power grid is in general available in urban areas, while off-grid energy sources,
coming from renewable ones, should be exploited in rural and low-income areas.
Focusing instead on the monthly subscription fee that the users pay, a similar fee
could be paid by people living in urban and rural areas. However, people living in
low-income zones should pay low fees. As a result, the business model behind the
deployment of the network is based on ROI for the urban zones, while it needs to
be subsidized by the government in both rural and low-income areas. In any case,
the required network flexibility is always high. This is true in urban zones, where
the stringent requirements in terms of bandwidth and delay are translated into large
flexibility requests to the network. However, the network flexibility is also required
in rural and low/income zones, where the network should cope with the fact that
the energy coming from renewable energy sources is not always available. Finally,
Efficiency/Off Grid Operation 131
focusing on the user mobility, pedestrian, vehicular and high speed vehicular are
normally characterizing urban areas; while rural and low income zones may be
subject to lower mobility schemes compared to urban ones.
Given these challenges, we detail in Sec. 2 the main building blocks of an efficient
5G off-grid architecture. We then analyze the costs in Sec. 3 and face the costs
minimization in Sec. 4. Sec. 5 details the efficient management of an UAV-based
networks. Finally, Sec. 6 concludes the chapter.
2 An Efficient 5G Off-Grid Architecture
We describe in this section the main features which allow to define a 5G architecture
for rural and low income areas, based on the requirements in terms of efficiency and
off-grid operation analyzed in the previous section. Tab. 2 reports the main pillars
that should be followed in order to deploy such architecture. First of all, the required
level of flexibility imposes to deploy a converged solution, in which the operator
has full control of the network and the computing resources. Moreover, there is
not a strict separation between the devices at the edge, core and metro levels: each
device becomes a commodity, which is transparently used independently from its
level. A second aspect which allows a converged solution is the large exploitation
of functionalities which are completely virtualized. In this way, both the network
and the computing components are virtual functions which are controlled by an
orchestrator managed by the operator. In addition, the management of the virtual
components on a set of physical devices allows to improve the efficiency of the whole
architecture. Focused then on the available energy, we expect a large exploitation
of the energy from the sun, and hence the deployment of solar panels to power
the physical devices. Clearly, this solution needs to be coupled with a set of backup
batteries to provide the required level of energy when the sun is not available. Finally,
focusing on the technology options, we expect the adoption of two distinct solutions,
namely: i) Unmanned Aerial Vehicles (UAVs) carrying radio network elements, and
in particular Base Stations (BSs) [2] and ii) Large Cells (LCs) mounted at ground
site, realizing massive antenna arrays covering vast portions of territory [3].
To give more insight, Fig. 1 reports a scheme of the energy sources feeding a
typical site hosting 5G equipment. In particular, three sources of energy are identified,
namely: i) the power grid, which may not be available for all the locations and/or
not reliable, ii) the solar panels providing energy from the sun, and iii) the backup
batteries which are fed by the solar panels when extra-energy is generated, and
supply energy when the generation is not enough. As shwon in Fig. 2, when the
power grid is either unreliable or unavailable, as in the case of disdvantaged areas,
emergency situations, or emerging countries, a backup traditional diesel generator
might be needed. Power supply through diesel generator is already quite common
in emerging countries in which the demand of communication services, as well as
the demand of electricity in general, is growing at a faster pace than the evolution
Table 2 Main Pillars of the Efficient 5G Off-Grid Architecture [1]

Pillar Description
The networking and computing resources are jointly managed by an orchestrator.
Converged Solution
The physical devices of the access network are managed in conjunction with the
metro and core ones.
Virtualization of network and computing components by means of virtual functions
Virtualization of Network Components that are controlled by a centralized orchestrator.
Efficient management of the virtual resources on a set of physical devices.
Exploitation of general purpose HW to host the virtual functions in order to reduce
Exploitation of Commodity Hardware
CAPEX and OPEX costs.
Massive exploitation of solar panels to power the physical devices.
Solar-Powered Energy-Efficient Devices
Exploitation of backup batteries to provide electricity when the energy of the sun is
not available.
Exploitation of the UAVs to carry radio network elements.
Unmanned Aerial Vehicles and Ultra-Large Cells
Exploitation of LC mounted at ground to realize massive antenna array covering
ultra-large sizes.
PV PANELS POWER GRID
STORAGE
Fig. 1 Power supply for a 5G site with solar panels.
of the power grid. Frequent power outage events are possible in these environments,
especially in periods of peak of electricity demand.
Focusing instead on the 5G sites, Fig. 3 reports the LC-based and UAV-based
deployment options. In the LC-based solution (Fig. 3.(a)) the 5G site hosts the
whole radio functionalities, which are installed in Commodity Hardware (CHW)
and Dedicated Hardware (DHW). While we expect that most of functionalities are
run by CHW, specific ones, like the processing of low-level functions, will be run in
the DHW part of the site. On the other hand, in the UAV-based solution (Fig. 3.(a))
there is a physical splitting between the CHW (which is left at ground) and the DHW
(which is carried by the UAV). In this way, it is possible to limit the weight of the
load carried by the UAV, with a positive impact of the UAV flight time. However, the
physical splitting requires to consider the physical channel constraints between the
PV PANELS POWER GRID
DIESEL
STORAGE GENERATOR
Fig. 2 Power supply for a 5G site with solar panels and diesel generator, in case the power grid is
not available.
CHW and the DHW, which include e.g., a maximum distance that has to be enforced
between the UAV and the site, and Line Of Sight (LOS) conditions.
DHW
CHW
Ground Site
Coverage Area
(a) LC-Based
UAV
DHW Radio Link
DHW
CHW
Coverage Area Ground Site
(b) UAV-Based
Fig. 3 5G deployment options under consideration [4].

3 Cost Analysis
Up to this point, a natural question is: Is a 5G architecture tailored to off-grid

scenarios efficient from a cost perspective? To answer this question, we consider
the LC-based and UAV-based solutions reported in the previous section, and the
set of parameters from [5], which include: the site cost, the battery cost, the solar
panel cost, the CHW cost, the DHW cost and the UAV cost. We then apply the
methodology of [5] to estimate the number of 5G sites and the required number
of batteries and solar panels (which we assume are the same for all the sites), by
considering three representative scenarios in the world, namely: i) the area of Frascati
(Italy scenario), ii) the Rarotonga island (Cook Islands scenario), and iii) the city of
Harare (Zimbabwe scenario). For each scenario, we consider the LC-based and the
UAV-based solutions.
Fig. 4(a) reports the outcomes in terms CAPital EXpenditures (CAPEX) for the
different scenarios and the different solutions. Interestingly, we can note that the site
cost generally dominates over the other ones in the Italy and Cook Island scenarios. In
addition, in the Zimbabwe scenario the site costs are lower compared to the Italy and
Cook Islands ones. Moreover, the cost of solar panels and batteries is always much
lower compared to the other costs. Eventually, the UAV-based solution includes also
the cost for buying the UAV, which is again pretty low compared to the other ones.
Overall, the total CAPEX costs are in the order of millions, being the Zimbabwe
scenario the least expensive for the operator.
In the following, we investigate the impact of introducing OPerating EXpenditures
(OPEX) costs and revenues from users in our analysis. Focusing on the OPEX costs,
we introduce the maintenance operations. In addition, we assume that the revenues
from users are computed from a given monthly subscription fee, which is applied
to all the users. Moreover, for the Italian and the Zimbabwe scenarios, we assume
that the energy can be derived also from the electricity grid. On the other hand, no
connection from the electricity grid is assumed in the Cook Island scenario. Focusing
instead on the revenues, we introduce the monthly subscription fee from users, by
adopting the conservative assumption that all the users of a given operator pay the
same fee. Fig. 4(b) reports the obtained results in terms of Net Present Value (NPV),
an economic indicator including CAPEX, OPEX, revenues from users. In particular,
when the NPV assumes positive values, the 5G deployment becomes profitable by the
operator. The different curves of Fig. 4(b) are obtained by varying the user monthly
subscription fee for the different deployment options and the different operators.
Interestingly, we can note that, even by setting very low monthly subscription fees,
in the order of 1 [EUR] for the Zimbabwe scenario and 10 [EUR] for the Cook
Island and Italian ones, it is possible to generate a profit for the operator. Overall,
the considered architecture is feasible, even in the Cook Island scenario, where the
lack of the electricity grid forces the installation of a large number of batteries and
solar panels.
6
x 10
10
Batteries
Solar Panels
8 Commodity HW
Dedicated HW
UAV
CAPEX [Euro]
6 Site Acquisition
0
UAV − Italy LC − Italy UAV − Cook Islands LC − Cook Islands UAV − Zimbabwe LC − Zimbabwe
Strategy − Scenario
(a) CAPEX Breakdown

9
10
UAV − Italy
LC − Italy
8 UAV − Cook Islands
10
LC − Cook Islands
NPV [Euro]
UAV − Zimbabwe
10
7 LC − Zimbabwe
6
10
5
10 −1 0 1 2
10 10 10 10
Monthly Subscription Fee [Euro]
(b) NPV, 5% Discount Rate
Fig. 4 Capital Expenditure (CAPEX) breakdown and Net Present Value (NPV) by applying the
UAV-based and LC-based strategies over the considered scenarios [5].
4 Cost Minimization
In the following, we then target the problem of cost minimization for the off-grid 5G
architecture, by focusing on the UAV-based solution. The problem is then sketched
as follows. Given: a set of UAVs, a set of areas to be covered, a set of candidate
5G sites and a set of candidate physical links connecting the sites. Minimize: total
CAPEX cost, including: solar panels, batteries, 5G sites, physical links. Subject to:
i) coverage of the areas, maximum number of batteries/solar panels per site, amount
of energy available at the 5G site, amount of battery available on the UAV. The
problem is then mathematically formulated and solved in [6], while here we report
the main outcomes. In particular, we focus on a zone in Frascati (Rome), with the
set of parameters reported in [6].
The optimal solution selects 3 sites to install out from a list of 9 candidates. Tab. 3
reports the breakdown of the number of batteries and solar panels installed in each
site. Interestingly, we can see that each site has a specific set of solar panels/batteries.
By further investigating this issue, we have found that each site manages a specific
set of UAVs, which results then in a variegate amount of energy demanded to each
site.
Fig. 5 reports then the breakdown of the total costs for the optimized UAV-based
solution. Interestingly, the site acquisition costs dominates over the other ones, which
Table 3 Breakdown of the installed SPs and batteries for the UAV-based solution [6].
Metric Site ID
9 10 15
Batteries 21 15 15
Solar Panels 10 8 7
Cs=40.0
40
30
Cost [kEuro]
20
10
CD=4.3
CSP=0.8 C =0.15
B
0
Site UAV SP Batt.
Equipment Type
Fig. 5 Breakdown of the equipment costs for the UAV-based solution [6].
6
x 10
4.633
Centers of the Areas

4.6325
Coverage Boundaries
[m]
Covered Areas
4.632 Installed Sites and Links
4.6315
3.065 3.07 3.075 3.08 3.085
5
[m] x 10
Fig. 6 Installed sites and fiber links for the UAV-based solution [6].
include the solar panels, the batteries, and the UAVs. This outcome confirms our
previous finding of Fig. 4(a): the site acquisition costs are important aspects that
should be considered during the deployment of 5G networks in off-grid scenarios.
Fig. 6 reports the installed sites and physical links in the considered scenario.
Differently from currently deployed cellular networks, in our considered 5G archi-
tecture it is not mandatory to install a site in each area. In the UAV-based solution,
in fact, the BS functionalities are carried by the UAV, thus allowing different zones
to be covered solely by the UAV, and hence limiting the total costs.
Finally, Fig. 7 reports the total battery level in each installed 5G site. We consider
a time period equal to one month in our analysis. Interestingly, we can note that the
battery level presents a clear day-night trend, with an increase during the day (as a
consequence of the fact tha the solar panels are able to recharge the batteries), and a
decrease during the day. Each site has a specific maximum battery level, due to the
fact that the number of installed batteries is not the same across all the installed sites
4
10 Site 9
6
Site 10
Site 15
5
Battery Level [Wh]

4
1
100 200 300 400 500 600 700
TS Index
Fig. 7 Temporal variation of the battery levels [6].
2. Move
3. Cover
4. Move
1. Recharge
5G Site
Covered Area
Fig. 8 An example of an UAV mission composed of recharging, moving and covering actions [7].
(as reported in Tab. 3). Finally, even considering the same site, we can note that the
battery level does not have always the same trend: this is due to the fact that each
UAV can move also across the sites, and hence varying the total demand of energy
required to each site.
5 Efficient Management
The efficiency of the 5G network can be further improve by minimizing the amount
of energy consumed during operation. In the considered UAV-based architecture, this
is achieved by minimizing the amount of energy used by the UAVs when covering
a set of areas. Fig. 8 reports an example of an UAV mission composed of different
steps, namely: i) recharging in a ground site, ii) moving from the ground site to an
area, iii) covering of an area, and iv) moving back to the ground site. In order to limit
the amount of energy requested to the ground site during the recharging operation, it
is of mandatory importance to reduce the energy consumed when moving the UAV.
Therefore, an efficient management of the UAVs is pursued in this step. In order to
target this goal, we introduce a framework based on multi-period graphs, a powerful
modeling tool which allows to model the trajectory of UAV through space and time.
REC COV
MOV
MOV
Site 1 Area 1
STAY
MOV MOV
MOV MOV
MOV MOV
Site 2 Area 2
MOV
STAY
MOV
REC COV
Fig. 9 Possible links between two sites and two areas. Each link consumes one TS [7].
To this aim, we associate a link to each possible action, namely: covering an area,
moving, and recharging. Fig. 9 reports a scheme of the links between two sites and
two areas. The idea of the multi-period graph is to associate a time slot to each
arc, and to consider the pairs (place,time slot) as nodes of the graph, and the links
as the UAV action performed between one node and another one. Fig. 10 reports a
generic multi-period graph composed of one source node, different nodes, and one
one sink node. Each link is an action performed by the UAV. When the place is not
varied between one time slot and the following one, the following actions can be
performed: i) recharging if the current place is a site, ii) staying fixed at ground (and
not consuming energy) if the place is again a site, iii) covering if the current place
is an area. On the other hand, when the place is varied, the only possible action is
moving between the two places. We then associate a flow variable to each UAV and
to each link. By properly setting the flow variables over the multi-period graph, we
are able to control the energy consumption and the trajectory of the UAVs (we refer
the reader to [7] for a detailed explanation on this aspect).
We define the following energy-efficient (EE) strategy over the multi-period graph.
Given: a set of areas to be covered, a set of UAVs, a set of sites, a set of solar panels
and batteries for each site. Minimize: total energy due to moving of the UAVs.
Subject to: energy available at the 5G site, coverage of the areas, minimum and
STAY STAY
REC REC
COV COV
p1, 1 p1, (t − 1) p1, t p1, (t + 1) p1, T
Ω
Υ MOV MOV
Υ Ω
Υ MOV MOV
Ω
p2, 1 p2, (t − 1) p2, t p2, (t + 1) p2, T
COV COV
REC REC
STAY COV
Fig. 10 General transitions between the states of the multi-period graph. The source node Υ, a set
of two places {p1, p2 }, and the final sink node Ω are shown [7].
Table 4 EE and MC comparison [7].
EE
Metric MC
TL=2 [h] TL=24 [h] TL=48 [h]
Energy due to moving operations [Wh] 86982.4 31150.6 29115.7 23858.8
Problem gap [%] < 10−6 54.6 50.2 38.1
maximum battery level of each UAV. We then consider a rural area in Frascati, and
a representative set of parameters, reported in [7]. We then run the energy-efficient
(EE) strategy based on the multi-period graph, and compare it against a maximum
coverage solution (MC), which does not take into account the energy consumed by
the UAV. Tab. 4 reports the comparison between the two strategies, by considering
the EE solution obtained after 2, 24 and 48 [hours] of computation. We consider as
terms of comparison the total energy consumed by the UAV for moving operations
and the values of problem gap of the obtained solution w.r.t. the optimal one. Several
considerations hold in this case. First, the energy due to moving is notably reduced by
the EE strategy, which ensures full coverage like the MC one. Second, the energy due
to moving tends to decrease as the computation time of the EE problem is increased.
Third, the gap of the MC solution is very low, as this solution is pretty easy to be
retrieved. Fourth, the gap of the EE strategy is higher than the one of MC, as the
former is much more complex to be solved compared to the latter. Nevertheless, the
EE strategy is able to find a solution already limiting the values of energy consumed
by the UAV even after two hours of computation. Overall, the presented results
demonstrate that it is possible to efficiently limit the amount of energy consumed in
an UAV-based 5G network based on an off-grid architecture.
FEBRUARY JULY DECEMBER

800
RE power [W]
Days Days Days

600
Mean Mean Mean
400
200
0
5 a 7 a 9 a 11 1 p 3 p 5 p 7 p 9 p 5 a 7 a 9 a 11 1 p 3 p 5 p 7 p 9 p 5 a 7 a 9 a 11 1 p 3 p 5 p 7 p 9
m m m am m m m m m m m m am m m m m m m m m am m m m m
Daytime [hour] Daytime [hour] Daytime [hour]
Fig. 11 Daily profiles of renewable energy production per 1 kW p in different sample months in
Torino (results obtained with PVWatts) [9].
5.1 Renewable Energy Sources
An interesting option to power mobile networks are renewable energy sources (RES).
Besides the potential benefit in terms of OPEX reduction, the introduction of RES is
particularly relevant when we consider the need to bring cellular network services to
portions of the world population that do not have access to a reliable power grid. The
demand of cellular networks and services is growing fast in geographical areas in
which the power grid is not reliable and long periods of power outages are frequent.
The solution most frequently adopted by operators relies on the use of diesel power
generators, which are however extremely costly, because of the price of fuel, the cost
to transport it and the frequent costly maintenance interventions. In these scenarios,
the use of RES has become an extremely attractive option [8]. The typical power
supply in this case is similar to the one reported in Fig. 2, with small solar panels
that power a BS or a few BSs, some battery units where extra energy can be stored
to be used when needed, and possibly a backup generator, such as a diesel generator.
Even in geographical areas in which the power grid is available and reliable,
the solar solution can be interesting: it can be economically effective in rural areas,
where the cost of bringing a power cable to the BS may be higher than that of a
solar panel and in urban environments, in which bringing a power connection to a
BS may require digging across streets. In addition, the introduction of RES allows
for a reduction of operational costs.
A RES-based power supply raises, however, some critical issues associated to the
intermittent nature of RES. As an example, Fig. 11 shows the daily energy production
profiles of a solar panel of 1 kWp (the kWp, kilowatt peak, is the maximum electric
power that can be supplied by a photovoltaic panel in standard conditions) in Torino
[9]. The plot on the left refers to the days of February, the one on the right to
0.05 0.2
Empt
Full
0 0
20 25 30 35 40 20 25 30 35 40 20 25 30 35
PV size [kWp] PV size [kWp] PV size [kWp]
(a) Average battery charge (b) Pe (c) Pf
Average battery charge, empty Efficiency/Off

(Pe ) and fullGrid(P f ) battery probabilities versus PV panel size, for different
Operation 141 time slots, for Turin, with 5
range discretization, for CB = 25 kWh, QS = 100 Wh, with residential weekday traffic profile, adpoting the Earth model.
Average battery charge [kWh]

30 30
All Days All Days All Days
Type 1 25 Type 1 25 Type 1
Type 2 Type 2 Type 2
Type 4 20 Type 4 20 Type 4
15 15
10 10
5 5
0 0
5 10 15 20 5 10 15 20 5 10 15 2
Hour Hour Hour
(a) Residential (20kWp) (b) Residential (30kWp)
Fig. 12 Average hourly battery charge for a panel of 30 kW peak, battery capacity equal to 25 kWh,
(c) Residential (40kWp)
in Torino [11].
30 30
All Days
Type 1
Battery charge [kWh]

25 are obtained with the tool PVWatts and derived from 25
Type 2 those of July; the values
Type 3 real location based data
Type 4 20 and considering the Typical Meteorological Year [10]. 20
Type 5 Some aspects of these patterns are critical for the design and dimensioning of the
power supply system. 15 First, there are several hours (during night) in which there is 15
no production at all. 10 Since service continuity is needed, powering the BS during 10
these periods require some power All supply
Daysthat is alternative
Type 3 to the RES, either the All Days Type 3
power grid or some battery5 that must Type Type 4 charged. Second, there 5
have1 been previously Type 1 Type 4
are quite significant differences in the Type 2
various Type 5In a location like Torino,
seasons. Type 2 Type 5
0 0
5 10 15 20 the summer production is 5
roughly 10
double the 15
winter 20
production. Hence, a power 5 10 15 2
Hour supply system that is dimensioned over theHour winter electricity demand of the BS is Hour
(d) Business (20kWp) over-dimensioned during summer (e) Business (30kWp)
time. Finally, even in the same season, there are (f) Business (40kWp)
differences in the energy production levels among different days. This implies that
Average hourly battery charge for a macro
a proper BS in Turin
dimensioning needsversus
to copetime, witheffects
with the 5 and production
NW of=variable equal-range discretization, time slot T = 1h,
levels
0, and 40 kW peak, CB = 25 kWh, and QS = 100 Wh, for both the residential and business weekday traffic profiles, adpoting the Ea
even in the short term.
To understand the effect of production level variability among days of the same
season, consider the case analyzed in [11]: a winter period in the city of Torino, a
power supply system of 30 kWp, and a battery with 25 kWh capacity. We partition
Granularity which level,
the winter days into five types based on their production T =6 h type
so that and1 collects
NT = 4; (iii) Hourly model, M
the 20% of the days with the lowest production,Tand=1 type h5 collects
and NT = of24;
the 20% the (iv) 30-minute model, M
we investigate the impact daysof the
the choice of the and time
the dayhfor and 48; and (v) 15-minute mod
with highest production, types 2-4 collect intermediate cases. Fig. 12
T of=0.5 NTday=types.
ty, i.e., of the value ofshows thetheparameter T , which
average battery charge in different hours the five
The thick blue line is the average. During night, withthe battery
T =0.25 h toand
is used NTthe= 96. We discuss the effec
power
s the time slot in the DTMC model. Wetheconsider
results to be a granularity
BS (assumed for deriving on the data for Turin. Same conclusions
LTE BS). The battery charge reduces,
s: (i) Daily model, MD in short,
almost in which
at constant T night.
rate, during h
=24 During day time, the solar production allows
thedaycase
of typeof Paris.
= 1; (ii) Periods of day model, M can in
is large, the battery P
short, in
to power the BS and to recharge the battery. In 5, the energy production
hence be fully charged during the day. On the contrary, in
days of type 1 the production is low and the battery is charged only partially. A few
consecutive days of type 1 will hence induce some BS power outage, unless some
backup supply is available. The results suggest that, when dimensioning the power
system, a careful evaluation of the effects of energy production variability is needed.
In traditional 3G and 4G cellular systems, the power demand of the BSs is very
little load proportional, i.e., it is almost constant regardless the load that is carried.
This does not cope well with the characteristics of the RES power supply, that is
highly variable. Indeed, it turns out that, for an entirely off-grid operation, RES
power supply systems, solar panels and batteries, have to be quite over-dimensioned,
so as to provide power even in low production seasons, and in case of consecutive
days of little production. To this extent, the flexibility of 5G technology brings new
opportunities. The possibility to allocate resources in a very dynamic way, and to
dynamically distribute the virtualized network functions among the nodes of the
network, translates into the possibility to make also the electricity demand more
flexible and more suited to be powered with RES.
6 Conclusions and Future Work
In this chapter, we have faced the aspects of efficiency and off-grid operation of
a 5G network tailored to serve rural and low-income areas. After proposing a set
of pillars of a reference architecture, we have analyzed the total costs that are
incurred when UAV-based solutions and LC-based ones are deployed. Overall, the
considered solutions allow to achieve a profit for the operator, even when the monthly
subscription fee is in the order of 10 [EUR] for the rural scenarios and 1 [EUR] for the
low income ones. Moreover, we have shown that the site acquisition costs generally
dominate over the other ones, which include e.g., the solar panels, the batteries, and
the UAVs. In the second part of our work, we have targeted the minimization of the
CAPEX costs, by showing that an UAV-based solution is able to limit the number
of ground sites used to host 5G equipment, while allowing the coverage of the areas
through the exploitation of the UAVs. Finally, we have targeted the energy-efficient
operation of a 5G network, by adopting a framework based on multi-period graphs
in order to model the trajectory of the UAVs, as well as the actions that they take over
time. Results show that it is possible to wisely reduce the amount of energy spent
for moving the UAVs, while allowing the coverage of the areas.
As next step, we plan to face several implementation aspects, such as: the presence
of regulatory constraints which limit the UAV flight over the users, the practical
limitations introduced by the radio channel between the UAV and the ground site,
the definition of the technologies that are exploited for the realization of the LCs,
the modelling of the energy of the UAV by considering also the height at which the
UAV are required to fly.
Finally, we have considered the introduction of renewable energy sources to
power the nodes of the network. The case is interesting in a number of scenarios: in
emerging areas in which the power grid is not reliable, in emergency situations or
disadvantaged areas in which bringing the power is hard and costly, in other cases as
a mean to reduce operational costs. Renewable sources are intermittent and highly
variable, both on the time scale of days as well as on a seasonal basis. While in a
traditional network the power requirements are extremely rigid, in the 5G case the
network is more flexible and it is, hence, possible to envision a network that can
easily and effectively adapt to the highly variable patterns of energy production that
are typical of renewable sources.
Acknowledgements This work has received funding from the project BRIGHT, under the Call
Mission Sustainability 2016.
References
1. L. Chiaraviglio, N. Blefari-Melazzi, W. Liu, J. A. Gutierrez, J. Van De Beek, R. Birke, L. Chen,

F. Idzikowski, D. Kilper, J. P. Monti, et al., “5g in rural and low-income areas: Are we ready?,”
in ITU Kaleidoscope: ICTs for a Sustainable World (ITU WT), 2016, pp. 1–8, IEEE, 2016.
2. M. Mozaffari, W. Saad, M. Bennis, and M. Debbah, “Drone small cells in the clouds: Design,
deployment and performance analysis,” in Global Communications Conference (GLOBE-
COM), 2015 IEEE, pp. 1–6, IEEE, 2015.
3. M. Eriksson and J. van de Beek, “Is anyone out there? 5g, rural coverage and the next 1 billion,”
IEEE ComSoc Technology News (CTN), 2015.
4. L. Chiaraviglio, W. Liu, J. A. Gutierrez, and N. Blefari-Melazzi, “Optimal pricing strategy
for 5g in rural areas with unmanned aerial vehicles and large cells,” in Telecommunication
Networks and Applications Conference (ITNAC), 2017 27th International, pp. 1–7, IEEE,
2017.
5. L. Chiaraviglio, N. Blefari-Melazzi, W. Liu, J. A. Gutiérrez, J. van de Beek, R. Birke, L. Chen,
F. Idzikowski, D. Kilper, P. Monti, et al., “Bringing 5g into rural and low-income areas: Is it
feasible?,” IEEE Communications Standards Magazine, vol. 1, no. 3, pp. 50–57, 2017.
6. L. Chiaraviglio, L. Amorosi, N. Blefari-Melazzi, P. DellâĂŹOlmo, C. Natalino, and P. Monti,
“Optimal design of 5g networks in rural zones with uavs, optical rings, solar panels and
batteries,” in 2018 20th International Conference on Transparent Optical Networks (ICTON),
pp. 1–4, IEEE, 2018.
7. L. Amorosi, L. Chiaraviglio, F. D’Andreagiovanni, and N. Blefari-Melazzi, “Energy-efficient
mission planning of uavs for 5g coverage in rural zones,” in 2018 IEEE International Confer-
ence on Environmental Engineering (EE), pp. 1–9, IEEE, 2018.
8. V. Chamola and B. Sikdar, “Solar powered cellular base stations: current scenario, issues and
proposed solutions,” IEEE Communications Magazine, vol. 54, no. 5, pp. 108–114, 2016.
9. D. Renga and M. Meo, “Modeling renewable energy production for base stations power
supply,” in Proceedings of the IEEE International Conference on Smart Grid Communications
(SmartGridComm), November 2016.
10. A. P. Dobos, PVWatts Version 5 Manual. Available at http://www.osti.gov/scitech/
servlets/purl/1158421.
11. A. C. da Silva, D. Renga, M. Meo, and M. A. Marsan, “The impact of quantization on the design
of solar power systems for cellular base stations,” IEEE Transaction on Green Communications
and Networking, vol. 2, no. 1, pp. 260–278, 2018.
5G security
Giuseppe Bianchi
Abstract Besides significantly outperforming past generations in terms of capacity

and throughput, 5G networks and systems will provide an infrastructure for the sup-
port of highly diversified and heterogeneous services. Indeed, many heterogeneous
application contexts will be profoundly impacted by 5G systems - to mention a
few: Industrial Internet and smart control systems, autonomous vehicles and drones,
life-critical e-health and remote surgery, virtual reality and augmented reality, re-
mote diagnostic and preventive maintenance, and so on. Service customization and
ultra-rapid deployment will leverage a virtualized network infrastructure, flexibly
integrating software-based network functions in both the network core, as well
as relocating time critical and low latency processing tasks down to the network
edge. Such a diversified and heterogeneous scenario calls for radically new security
models, capable to overcome the “one-size-fits-all” approach to security that has
characterized cellular systems until the latest fourth generation. In this chapter, after
a brief excursus on how security has evolved throughout the various generations of
cellular systems, we will focus on the security vision in 5G, and on the relevant major
approaches and challenges, including the discussion of novel threats to virtualized
systems.
1 Security in pre-5G systems
The problem of security in cellular systems has arisen initially to solve a very specific
problem: how to authenticate users connecting to the network, and protect the rele-
vant data in transit from attackers able to eavesdrop the radio channel. This activity
has been duly addressed during the previous generations of cellular systems, with
Giuseppe Bianchi
Universitá degli studi di Roma Tor Vergata, Dipartimento di Ingegneria Elettronica
Via del Politecnico 1, 00133, Roma
e-mail: giuseppe.bianchi@uniroma2.it
145
146 Giuseppe Bianchi
solutions that, although gradually, have today reached a level of protection consid-
ered completely satisfactory, to the extent that it is hard to find any breakthrough
improvement in this area in the last fifteen years.
The (in)Security of 1G and 2G systems

While the first generation systems did not specify any solution for the protection
of communications, GSM (i.e., the second generation) began to explicitly introduce
solutions for user authentication and encryption at the radio interface level. However,
GSM security solutions have proven to be extremely preliminary and insufficient for
many reasons, starting from the total inadequacy of the cryptographic techniques
adopted, and the nefarious “security by obscurity” strategy employed. Specifically,
the cryptographic algorithm adopted in the GSM authentication handshake, later on
called COMP-128, had not been validated by the cryptographic community, but had
been deliberately secreted, with the idea - a posteriori proven disastrous - that the
secrecy of the algorithm itself could increase security. Unfortunately, this was not
the case. As soon as, around 1998, the details of the COMP-128 algorithm leaked
out, it was a matter of a a few weeks for the cryptographic community to badly break
it, and prove its complete inadequacy!
Still, the bad experience with GSM taught the security community a very impor-
tant twofold lesson. First, it is not nearly easy to prevent leakage when an highly
critical algorithm must be implemented by numerous actors (operators, SIM manu-
facturers, authentication center producers, etc), and when the devices that implement
it can be subject to “reverse engineering” of the related software code. Second, it
is extremely unlikely that an algorithm not specifically developed by professional
cryptographers, nor duly analyzed by the cryptographic community, results to be
sufficiently robust.
Although the cryptographic algorithm is the most striking weakness of second
generation systems, the security shortcomings of GSM were not limited to this. In
particular, GSM did not provide mutual authentication. In fact, in GSM systems,
while the user needed to undergo authentication before being permitted to access the
network, the opposite was not true, i.e., the user was not provided and means to verify
the authenticity of the radio station to which he was attacking to. This feature was
perhaps not perceived as critical during the early years when the GSM system was
being standardized. But the technological evolution occurred during the late ‘90s,
with the emergence of programmable radio devices (Software-Defined Radio), has
made not only possible but even quite cheap attacks based on “rogue base stations”,
i.e., fictitious radio stations controlled by an attacker, which was thus made able to
intercept and tamper with the end users’ communications.
Finally, no security solution in the core network part had been standardized in
GSM. The encryption on the radio interface terminated in the access network; there-
fore the information was transported in the clear on the fixed network, with the result
that any attacker able to access the transport infrastructure was able to violate the
confidentiality and integrity of the data transported.
5G security 147
Fig. 1 4G security architecture: roman numbers on top of each arrow specify which security domain
is involved for the considered interface.
3G: the security generation.

The next third generation, UMTS, was probably the generation in which the greatest
progress has been made on security, both in terms of the quality of the solutions
adopted and in terms of interventions in the various sub-systems involved. Firstly, 3G
systems have completely abandoned the “security by obscurity”, by adopted publicly
scrutinized cryptographic algorithms in the AES (Advanced Encryption Standard)
family, algorithms which are much safer than previous ones, and in part are still
today state of the art. The application of cryptographic techniques has also been
significantly improved, both through explicit differentiation of encryption ciphers
and relevant keys from data integrity, and through the introduction of privacy and
protection features for users against attacks devised to recognize and track the end
user position (location privacy). 3G systems have also obviated the problem of rogue
base stations, by providing an extremely effective technique of mutual authentication.
Finally, 5G systems have duly addressed security in the netwpork core by providing
uthentication and protection mechanisms for both data transport in the cose as well
as protection of signalling.
Security systematization and 4G.

In line with the progress made in 3G systems, the fourth generation has made
several improvements, and has, above all, clearly inserted the theme of “security by
design”, i.e., by addressing security since the very beginning of the LTE architectural
specification phase. To this purpose, the overall 4G security architecture (Figure 1)
has been organized in five explicit domains:
I Network access security: protection at the level of air interface and secure
access to the service by the user;
II Network domain security: protect network elements and relevant exchange of

data traffic and signalling messages;
III User domain security: protection of the mobile terminal and its interfacing
with the USIM and device;
IV Application domain security: secure communications at the application layer;
V Visibility and configuration of security: means to permit to check if (and
which) security features are in operation, and how they are configured.
In addition, 4G systems brough about many punctual improvements (and when-
ever applicable also fixes) in the specific algorithms and techniques employed. These
improvements range from better authentication and key management, improved cryp-
tographic algorithms (including support for a new stream cipher called ZUC), end-
to-end security, integration with IP security technologies, etc. As such, they are quite
technical and incremental; we will see in the next section that a much more radical
revision of the security model is expected in 5G systems.
2 5G security: vision
As briefly summarized in the previous section, the past generations of networks

and cellular systems have made huge steps in the security sector, both in terms
of cryptographic algorithms (starting from 3G systems) as well as in fostering a
“security by design” approach and a relevant systematic organization of the security
tasks into five security domains. Given these advanced results attained by previous
generations, a question stands out: should security still play an important role in 5G
networks? Or the bulk of the security work has already been done, and 5G security
is expected to move along very detailed and quite incremental steps?
Although not officially included in any standard document, several 5G stakehold-
ers and alliances converge upon a 5G security vision revolving around three major
driving principles:
• Flexible security;
• Supreme built-in security;
• Automation.
The first item, flexible security, is in our opinion by far the most significant change
that 5G systems are called to address. Indeed, it is a direct consequence of the radical
change in perspective that characterizes the new generation of wireless networks.
While the previous 2/3/4G systems were specified for a well-identified class of users,
and had therefore defined a specific and unique family of security and data protection
solutions, the emerging 5G systems are born with the aim of supporting extremely
diversified vertical services, targeting different types of users, and including services
not exclusively dedicated to human users.
It follows that the “one-size-fits-all” paradigm, which had characterized security
solutions in previous generations, now becomes a questionable approach, which is
5G security 149
hardly applied to very diverse services encompassing heterogeneous technologies

and terminals with widely different capabilities and service requirements.
The “Flexible Security” paradigm promoted by 5G networks is devised to meet
such changed needs, by providing diversified security solutions, adaptable to the
specific (different) scenarios considered. For example, services with very low la-
tency requirements will need security solutions capable to meet such tight latency
constraints. And whenever the need for trade-offs emerge, service deployers should
be made able to decide whether to downgrade security requirements or performance
ones; in essence, 5G systems mandate for an increased flexibility in the ability of
the network to deploy the security technologies and solutions most appropriate to a
given scenario.
Some concrete initial steps towards flexible security have been made, with the
introduction of a new protocol, 5G EAP-AKA, specifically meant to support flexible
authentication in emerging 5G systems. Indeed, while authentication was based until
4G on a single and well-defined approach, diverse 5G services will have to take into
account numerous diversification factors. For instance, which terminal devices or
end users do we need to authenticate? And does the device come along with a SIM,
or it is a SIM-less terminal? And are there constraints and limitations to take into
account, such as energy consumption or computational capacity? And what level of
authentication and authorization must we provide to the different types of “mission
critical” services? The rationale of EAP-AKA is to act as “protocol container”, so
as to permit support for more tailored authentication handshakes to be deployed in
specific service contexts.
As for the other two points mentioned above, the vision underlying the idea of
having a “built-in security”, i.e., directly integrated since the early specification
stages of the emerging 5G network, is arguably not completely new (4G systems
fostered a similar vision), but it remains a crucial and fundamental vision. Moreover,
4G systems have left many challenges still widely open, and there are several specific
areas where the level of security of systems based on previous generations can be
significantly improved with solutions directly integrated into 5G systems. These
areas (at the very least!) include enhanced privacy, security assurance, and the need
to increase robustness against cyber-attacks owing to the alarming pace at which
malware is evolving on mobile devices.
Finally, in relation to the third point mentioned above - automation -, the flex-
ibility of security management must go hand in hand with tools to simplify the
management of security itself in the network, and allow a quick adaptation not only
of the operation of the network, but also of the security solutions in place, to new
emerging services.
3 Network virtualization and softwarization: implications on

security
A key feature of 5G systems is the migration towards a virtualized network infras-

tructure: network functions are no longer provided by dedicated hardware devices
or components, but are implemented in software and run inside virtual machines or
containers, which in turn reside in cloud architectures or, whenever necessary, are
relocated to the edge of the network so as to attain the low latency requirements
required by critical 5G applications.
In terms of security, the key advantage inherent in the virtualization of network
functions, and in the simplification in the related control and management offered by
the “Software Defined Networking” paradigm, consists in being able to on-demand
deploy security-related network functions, whenever and where they become neces-
sary. Such security-related functions include, but of course are not limited to: traffic
analysis modules, firewalls, intrusion detection and prevention systems, Deep Packet
Inspection systems, programmable and relocatable network probes for detection of
modern sophisticated intrusions (e.g. Advanced Persistent Threats make use of lat-
eral movements), and so on. Their re-implementation as dynamically deployable and
reconfigurable virtualized network functions (opposed to their implementation into
dedicated components or hardware modules) is expected to foster an unprecedented
level of flexibility in the management of security. Indeed, this new dynamic and
flexible management paradigm offers a significant opportunity to rethink security
functions following a “security-as-a-service” paradigm, as well as automate network
protection features.
To achieve this goal, however, scientific research is currently still engaged in
at least two fundamental challenges. Firstly, traffic analysis or packet inspection
modules, made of software running in virtual machines, can hardly achieve per-
formance comparable to those attained by dedicated hardware components, leaving
the implementation of scalable and high-performance modules as a top priority.
Secondly, a complete integration of security services in the network management
and orchestration platforms is still ongoing. Challenges include the identification
of platform-agnostic programming interfaces for heterogeneous security and mon-
itoring solutions, as well as compelling visualization of the security status of the
network through appropriate visualization techniques.
At the same time, the new paradigm of virtualization of network functions brings
about new problems and new challenges in terms of security, given the significant
expansion of the attack surface that the softwarization and programmability of net-
work functions entails. On the one hand, the ability to migrate network functions
as virtual machines leads to the need to identify (or adapt from the IT world) se-
cure authorization, management, migration and secure attestation/execution of SW
images that implement such virtualized network functions. On the other hand, the
physical separation of such functions in separate devices is no longer applicable as a
security measure, and it is necessary to think of separation and isolation solutions of
such functions in a virtualized context, including means to strengthen the isolation
5G security 151
for the “slices” that make up the network and for the functions responsible for such
segmentation, and means to provide independent security solutions for every “slice”
in which the network will be divided. However, as we will discuss specifically in
the next section, this activity is only apparently simple, and fundamental problems
recently emerging in the processor architectures used in virtualization infrastructures
may play havoc with the above isolation requirements.
4 Critical vulnerabilities in modern processors: implications on

isolation
The last few months have seen the emergence of a brand new wave of attacks,
which leverage fundamental performance optimizations tightly integrated inside the
architecture of modern processors. Such attacks are broadly referred to as “transient
execution attacks”, as they exploit the capability of processors to “anticipate” com-
putation (e.g., via branch prediction and out-of-order execution) so as to speed up
performance. The first two attacks of this family, namely Spectre1 and Meltdown2,
showed that exception or branch mis-prediction events might leave unauthorized data
in the CPU’s micro-architectural state, e.g., in low level caches.
What truly concerns the security community is that Spectre and Meltdown were
not isolated cases, but specific instances of a far more fundamental and general
new threat - a Pandora’s Box was opened! As a matter of fact, recent chronicles
report that, on August 14th of this year (2018), a new variant of such attacks called
Foreshadow was announced. Foreshadow had an even more disastrous impact than
previous attacks. First, it undermines the security and trust model based on Intel
SGX, a hardware-based trusted software attestation considered so far a fundamental
component in cloud security. Moreover, as we will discuss in more details later on,
a version of Foreshadow pointed out the emergence of critical vulnerabilities in
virtualization-based isolation. And such a nightmare is still ongoing: a few days ago,
on November 13, 2018, while we were writing this chapter, a systematic analysis of
this methodology of attack was published3. This analysis allowed not only to discover
seven brand new variants, but also showed that the problem is not limited to Intel
processors, but also extends to other vendors, namely ARM and AMD. Moreover,
this work shows that many of these new variants of attack do not appear mitigated
by the security patches issued so far by the processors’ vendors.
Transient execution attacks and virtualization

Our specific concern for 5G systems is that such new wave of transient execution
1 P. Kocher et. al., Spectre Attacks: Exploiting Speculative Execution, 40th IEEE Symposium on
Security and Privacy, 2019.
2 M. Lipp et. al., Meltdown: Reading Kernel Memory from User Space, 27th USENIX Security
Symposium, 2018
3 C. Canella et. al., A systematic evaluation of transient execution attacks and defenses, eprint
arXiv:1811.05441, Nov. 13, 2018, available at https://arxiv.org/abs/1811.05441
Cloud Server
Physical Core
Logical Core 1 Logical Core 2

Victim Attacker
VM VM
L1 Cache
L2 Cache
Victim’s memory Attacker’s memory
Fig. 2 Foreshadow-VMM attack: high-level overview
attacks may play havoc with the desire of virtual infrastructure operators to rely
on virtualization for network slicing’s isolation and segregation. A first threat to
isolation in virtualized systems was concretely shown by the Foreshadow4 attack.
Foreshadow, also known as L1 Terminal Fault, is a speculative execution attack that
provides the possibility to completely bypass the virtual memory abstraction, thus
providing means to read unauthorized data. For performance reasons, processors
use speculative execution during the virtual-to-physical memory address translation.
In particular, while the correspondence between virtual and physical addresses is
searched in the page table (i.e., during a page table walk), the processor accesses in
parallel the L1 data cache. If the logical address has not a mapping to the physical
location, the translation process is aborted and a terminal fault rises. However, there is
a time period before the retrieve operation in which data are still passed to the cache,
even if an access violation occurs, and thus where tailored side-channel methods,
similar to those used in the Spectre and Meltdown attacks, may be exploited to gather
access to protected information.
There are three different variants of the Foreshadow attack: (i) Foreshadow-
SGX, the first Foreshadow version, designed to infer data from SGX trusted exe-
cution environment; (ii) Foreshadow-OS/SMM affecting operating system, kernel
memory, and system management memory; (iii) Foreshadow-VMM affecting virtual
machines (VMs) and hypervisors (VMM). Our specific interest here is on the last
variant, Foreshadow-VMM, as it threatens virtualized environments (thus including
virtualized network scenarios), by allowing a malicious guest VM to read memory
belonging to the VM’s hypervisor, as illustrated in Figure 2.
Referring the reader to the original sources for most technical details, in extreme
summary the attack relies on the following facts. let us preliminarily recall that the
virtual memory exposes to the user is divided in chunks called pages, which are
4 O. Weisse et. al., Foreshadow-NG: Breaking the Virtual Memory Abstraction with Transient
Out-of-Order Execution, Technical report, 2018, available at https://foreshadowattack.eu/
5G security 153
then mapped to the physical memory using Page Tables which contains all pairs
of hvirtual_address, physical_addressi. For performance reasons, modern pro-
cessors maintain a cache of recently translated addresses, and forward the content
of the Page Table Entries directly to the cache control logic while simultaneously
performing relevant checks, thus including whether an entry is valid. This implies
that, even if a “not present” fault is raised, there is a relatively small period of
time in which the processor will continue to speculate on the validity of the data
stored in the L1 cache (and thus forward such cached data to the relevant processing
instructions), until the fault takes effect. In this period of time, a side-channel attack
similar in concept to Meltdown can be therefore used to read data from any arbitrary
physical address, as long as that address is currently loaded in the L1 data cache
and a not present page table entry is triggered for that address. In practice, such an
attack becomes feasible as long as the attacker succeeds in deploying a malicious
VM on the very same core of a chosen victim’s VM. Indeed, once this is done, since
the malicious VM (as any other VM) controls its own virtual-to-physical address
translation, and since this information goes directly inside the L1 cache, it suffices to
instruct a malicious VM with an offending kernel module5 to modify the page table
entry of its own page table so as to include any desired physical address, pass this
table entry to the L1 cache, and trigger a terminal fault.
Implications on 5G systems
The emergence of such a new generation of transient execution attacks should
loudly alert the 5G community about the potential threats which may affect high-
level segmentation and slicing techniques, i.e., techniques which target isolation
while remaining agnostic to low level details. As discussed above, even if logically
isolated, a virtual machine which shares a same CPU micro-architectural state with
a victim’s VM (i.e., pinned to the same core) may exploit the shared physical cache
to get access to unauthorized information. We therefore believe that supplementary
attention should be posed on how to extend current slicing and isolation frameworks
with supplementary policies which specify further physical micro-isolation require-
ments - e.g. by preventing that system-critical VMs share the same hardware with
user-loaded VMs, or that VMs belonging to two different slides share the same core.
This is especially important when deploying segmentation at the network edge, as
this is arguably the place where, on one side, low-end commodity hardware will be
exploited, and on the other side, non-necessarily trusted computing tasks provided by
end users will reside and might therefore attempt to interact with other VMs. Finally,
the above described attack has so far been proven only in a scenario comprising
virtual machines running on an hypervisor, while Network Function Virtualization
is nowadays moving towards more performing approaches, such as containers or
unikernels. Still, we do not nearly expect such technologies to be exempt from such
types of attacks.
5 The feasibility of such a solution was concretely tested by our own group - we are currently
looking at simpler (user-space) techniques not even requiring the injection of VMs with a modified
kernel.
5 Conclusions
The goal of this chapter was to raise attention on the key aspects and challenges
related to 5G security. Rather than giving a detailed, but probably boring, summary
of the punctual and/or incremental security improvements planned in the various
components and strata of the emerging 5G systems, we preferred to focus on a more
conceptual presentation of the new directions that 5G systems are expected to take,
and on the novel security issues that the fundamental 5G revolution in terms of
infrastructure’ softwarization and virtualization is raising.
Obviously, many supplementary security challenges are brought about by each
of the three specific innovative service scenarios fostered by 5G, namely: Enhanced
Mobile Broadband, Ultra-reliable and low latency communications, and Massive
Internet of Things (or machine type) communications. For reasons of space we have
limited to discuss how the “Flexible Security” paradigm promoted by 5G networks
is devised to meet the very diverse security needs of such services, as well as other
vertical ones. But it is worth to remark that, especially in the case of IoT, the ability of
an attacker to gather control of millions of devices (ability proven feasible by Mirai,
Bashlite, or Aidra - just to mention some first generation IoT botnets’ instances)
may lead to new attack scenarios of unprecedented scale and impact (just imagine a
massive IoT ransomware taking control of your homes and assets).
Last but not least, the very large number of stakeholders involved in vertical
services will require crucial attention to the privacy of the users and the controlled
disclosure of sensitive data. We advocate explicit rules and policies devised to
govern the way by means of which user profiling activities are conducted, and
users’ behavioral information is gathered and exploited to provide personalized and
customized service models.
Reliable Slicing in 5G Networks
Luca Valcarenghi, Alessio Giorgetti, Barbara Martini, Koteswararao Kondepu,

Molka Gharbaoui, Piero Castoldi
Abstract In 5G, the term slicing refers, in general, to the possibility for different
customers (usually called tenant) to share the same physical network. Thanks to
the softwarization of networks according to the Network Function Virtualization
(NFV) concept and the programmability of network connectivity through
Software Defined Networking (SDN), new network and service capabilities can
be envisioned by integrating networking, computing and storage resources while
serving a multitude of tenants. Each tenant is assigned a logical network that can
satisfy its requirements. Survivability is one of the most important requirements
especially for vertical applications requesting Ultra Reliable Low Latency
Communications (URLLC). In this chapter the concept of slice is introduced and
the some use cases for providing reliability in a slice.
1 5G Slicing
With the advent of NFV and SDN a novel network scenario is envisioned
enabled by network deployments into the cloud also extended to the network
edge and by programmability of network connectivity through network
controllers. This trend known as softwarization is enabling new unique network
and service capabilities by integrating networking, computing and storage
resources into one programmable and unified infrastructure while serving a
multitude of distributed smart devices and applications (e.g., robots, drones,
smart vehicles). As result, current communications network scenario is moving
156 Valcarenghi, Giorgetti, Martini, Kondepu, Gharbaoui, Castoldi
from having a separate network for each application (e.g., fixed telephone
network, mobile telephone networks, Internet access) to a single network shared
by different applications or verticals. Network Slicing is a key feature of the 5G
System that allows Operators to flexibly structure the network resources to
match the services offered to subscribers, third-party customers, including the
roaming scenario. The concept of slicing emerged as a way of setting up several
logical networks for different verticals on the same physical network. Each
vertical is then assigned to the logical network that guarantees the required QoS.
Such setup potentially allows communication providers to save capital and
operating expenditures (CAPEX and OPEX). However, as for any shared
medium, guaranteeing the required QoS to network slices sharing the same
physical network is not a trivial task and remains an open issue. In particular,
slice control and management planes shall be designed for slice provisioning and
dynamic reconfiguration and the data plane shall guarantee each slice
requirements (e.g., QoS requirements, slice isolation, etc.).
This chapter overviews how the concept of slice is defined in different Standard
Developing Organizations (SDOs) and research projects.
1.1 The Concept of Slice in SDOs and Research Project
Several SDOs are focusing on the network slicing concept 1. The Next
Generation Mobile Networks (NGMN) alliance defines a Network Slice
Instance (NSI) as “… a set of network functions, and resources to run these
network functions, forming a complete instantiated logical network to meet
certain network characteristics required by the Service Instance(s)” 2. In 2 the
network slicing concept consists of three layers depicted in Figure 1: Service
Instance Layer, Network Slice Instance Layer, and Resource layer. The Service
Instance Layer represents the services (i.e., end-user or business services) which
must be supported. The Network Slice Instance Layer provides the network slice
instances with specific network characteristics that are required by the related
Service Instances (e.g., Enhanced MBB, M2M, Enterprise and Industry). The
Resource Layer provides the physical or virtual resources for slice deployment.
Reliable Slicing in 5G Networks 157
Figure 1 Slice concept in 2
3GPP in TR 28.801 3 defines, in accordance with NGMN, a network slice

instance (NSI) as “… a set of network functions and the resources for these
network functions which are arranged and configured, forming a complete
logical network to meet certain network characteristics…”. In addition, it
defines the following phases of a network slice lifecycle: preparation phase,
instantiation, configuration and activation phase, run-time phase,
decommissioning phase. Moreover, it introduces three management functions to
manage the NSIs to support communication services: Communication Service
Management Function (CSMF), responsible for translating the communication
service related requirements to network slice related requirements; the Network
Slice Management Function (NSMF), responsible for management and
orchestration of NSI; and the Network Slice Subnet Management Function
(NSSMF), responsible for management and orchestration of a network slice
subnet instance (NSSI). Finally, it defines the different roles of the actors (e.g.,
costumers, providers, operators, etc.) involved in slice provisioning as depicted
in Figure 2. However, TR 28.801 does not specify how to implement such
functions and their relationship with respect to the ETSI NFV architectural
framework.
Figure 2 High level function of roles in 3
ETSI NFV EVE012 4 establishes the correspondence between a network slice

(3GPP) and a network service (ETSI NFV). There, ETSI describes that an NFV
Network Service (NFV-NS) can be regarded as a resource-centric view of a
network slice, for the cases where a NSI would contain at least one virtualized
network function. Moreover, ETSI NFV EVE012 proposes that 3GPP slice
management functions interact with ETSI NFV Architecture through the Os-Ma-
Nfvo reference point as depicted in Figure 3. However, in 4 it is stated that
“…3GPP slice-related management functions are still under definition in 3GPP
SA5 and future updates might require further analysis about the interaction
between 3GPP slicing related management functions and NFV-MANO…”.
Figure 3 Interaction between slice management functions (3GPP) and ETSI

NFV Architecture from 4
Additional definitions of slices have been proposed by the following

organizations:
 The Internet Engineering Task Force (IETF) in 5 6 7
 The Broadband Forum (BBF) in 8
 The Optical Networking Forum (ONF) in 9
 ITU-T through the Focus Group on IMT (International Mobile
Telecommunication)-2020 (FG IMT-2020) in 10.
 Metro Ethernet Forum (MEF) in [11][12]
Within the research community, the 5G-Transformer project 14 envisions three

functional layers for providing verticals with slices: a Vertical Slicer as the
logical entry point for verticals to support the creation of their respective
transport slices in a short time-scale (in the order of minutes), a Service
Orchestrator to orchestrate the federation of transport networking and computing
resources from multiple domains and manage their allocation to slices, and a
Mobile Transport and Computing Platform (5GT-MTP), that provides and

manages the virtual and physical IT and network resources on which slices are
deployed. Such architecture implementation in under development.
The 5G!Pagoda research project in 15 overviews most of the ongoing network

slicing-related activities. In addition, it proposes a functional architecture of the
slicing system and it delineates key implementation elements. The functional
architecture of the slicing system for the single domain is based on a Domain
Specific Slice Orchestrator (DSSO), a Slice Operations Support (SOS), a Slice
Management Plane (SMP), a Slice Software Layer (SSL), a Slice Resource
Layer (SRL), a Virtual Computing/Storage/Connectivity Infrastructure Layer
(VCSCI), and a Physical Computing/Storage/Connectivity Infrastructure Layer
(PCSCI).
Within the general architecture proposed by the research project SONATA 16, a
“Slice Management Functional Block” is defined and integrated in the NFV
Management and Orchestration (NFV MANO) functional block, detailed in the
ETSI NFV architectural framework. In addition, the implementation of such
functions has been developed within the project.
2 Reliability in 5G Slices
The envisioned 5G network architecture, including the Next Generation Core

(NG Core, i.e., the new Evolved Packet Core --- EPC --- for 5G) and the New
Radio Access Network (New RAN), will be heavily based on virtual network
functions (VNFs) [17]. Network function virtualization (NFV) enables an easy
introduction of new network services by adding dynamic programmability to
network devices (e.g., as routers, switches, and applications servers) that, in
turn, empowers fast, flexible, and dynamic deployment of new network and
management services. Moreover, network function virtualization also enables
network slicing by providing multiple instances of the same network function. In
this context, the dynamic service chaining allows the delivery of a new breed of
applications (e.g., cloud robotics, smart cities) by dynamically selecting and

composing computational and network services deployed as virtual functions
(VFs) in distributed micro-clouds located at the network Edge closer to the users
[25]. The exploitation of network function virtualization is foreseen also in the
NG core [18] and the New RAN technology [19]. In the NG Core, the different
network functions (e.g., Access and Mobility Function (AMF), Session
Management Function (SMF), Policy Control Function (PCF), Application
Function (AF), Authentication Server Function (AUSF), User Plane Function
(UPF), and User Data Management (UDM)) can be virtualized, as it has been
proposed for LTE-A [18], and placed in different virtual machines (VMs) or run
as a single bundle in one VM.
As specified in [19] “…Service continuity is not only a customer expectation, but

often a regulatory requirement …”. Thus, in a scenario where network functions
are virtualized, both hardware and software failures assume the same
importance, and their reliability shall be guaranteed. Similarly, reliability at
service chain level is important to assure proper service availability features to
application service platforms deployed by verticals [25][26].
2.1 Virtualised EPC reliability
In the technical specification document (TS) 23.007 [20], 3GPP specified

different failure detection and recovery mechanisms for EPC components,
including detection of path failure with the help of Echo Request/Echo Response
timer messages. Moreover, approaches for recovering failures in a scenario
where a mobile network function is virtualised can stem from schemes already
proposed for grid and cloud networking [21]. Furthermore, scalable architectures
for reliability management are being defined by ETSI NFV [22] and
implemented in current open source orchestration frameworks such as
Openstack [23]. However, the performance of resilience schemes based on the
aforementioned approaches once applied to 5G mobile network have not been
fully evaluated so far.
This section demonstrates the capability of recovering vEPC failures by means

of a vEPC in “hot backup”. Both working vEPC and backup vEPC are deployed
in multiple Network Function Virtual Infrastructure Points of Presence (NFVI-
PoPs) made available by the federated testbeds belonging to the SoftFIRE
project [24]. The demo is designed to evaluate the Service Recovery Time (SRT),
that is the time required to regain user equipment (UE) connectivity, when the
proposed resilient scheme is deployed in different NFVI-PoPs.
The considered scenario and the proposed resilience scheme are depicted in Fig.
4 and Fig. 5 by referring to functional elements of the Long Term Evolution-
Advanced (LTE-A) architecture.
The proposed resilience scheme considers a scenario where the vEPC fails (e.g.,
a virtual machine where the vEPC runs crashes). Fig. 4 shows the two
considered vEPC resilience schemes based on vEPCs hot backup deployed in
federated NFVI-PoPs. The one on the left features two co-located vEPCs (i.e.,
vEPCa and vEPCb deployed in Surrey 5GIC testbed) while the one on the right
features a remote hot backup vEPC (i.e., vEPCr) deployed in a different
compute resource available in another testbed (i.e., Fokus ). In the latter case
two testbeds will be contemporarily utilized to implement the resilience scheme.
In the Surrey 5GIC testbed two different VNF functions (vOAISIM and vEPC)
will be implemented by exploiting open source mobile platforms (i.e.,
OpenAirInterface-OAI). Here, vOAISIM VNF provides emulation of virtual
user equipment (vUE) and evolved NodeB (eNB) while vEPC will be used to
emulate the core network.
Fig. 5 shows the considered scenario and lifecycle event when vEPC VNF fails.
Here, when VNFs are deployed, vOAISIM connects with vEPCa, and Zabbix
server start monitors the VNFs that are associated corresponding Zabbix agent.
Note that each vEPC VNF and vOAISIM VNF deployement contain also Zabbix
agent. If the Zabbix server detects an anomaly activity in vEPCa (e.g., overload)
or does not receive any status report from vEPCa (i.e., vEPCa crashed) for a pre-
defined period of time (i.e., time to trigger the activity), the Zabbix server check
the status of the hot backup vEPC to initiate a recovery procedure. The
receovery procedure consist in reconfiguring vOAISIM to connect to the hot

backup vEPCr. Upon reconfiguration vOAISIM is able to communicate hot
backup vEPCr. Similarly, the experiment also demonstrate to the recovery based
on the local vEPCb deployed in 5GIC testbed.
Figure 4: RAN and Core network deployment in federated environment
Figure 5: Proposed scheme experimental evaluation setup
2.2 Service chaining reliability

The advent of SDN and NFV enables a convergent network-cloud ecosystem

offering more effective and operative network and service deployments on top
of virtual networking, computing and storage resources integrated into one
programmable and unified infrastructure while serving a multitude of distributed
smart devices and applications (e.g., robots, drones, smart vehicles) in turn being
part of the infrastructure itself.
Thanks to softwarization, a scenario can be envisioned where service providers

may offer not only communication services, but also virtualized computing and
storage capabilities by elastically slicing the (cloud and network) infrastructure
into partitions (i.e., network slices) offering customized network functions and
services (e.g., NAT, firewall, deep packet inspection) tailored for specific
applications. Moreover, with the softwarization of telecommunication
infrastructures, a new breed of applications can be conceived (e.g., cloud
robotics, smart cities) by dynamically composing (i.e., chaining) computational
and network services deployed as virtual functions (VFs) in distributed micro-
clouds located at the Edge of the current telecommunication infrastructure.
Indeed, SDN can effectively provide programming abstractions that can be
exploited for the dynamic enforcement and in-line steering of data traffic along
the network path of service chains (i.e., service chain paths).
Through slicing and dynamic service chaining, service providers can deploy
service infrastructures to serve many different verticals while saving capital and
operating expenditures (CAPEX and OPEX). However, the concurrent usage of
resources, the high dynamicity of services and the geographical distribution of
VFs pose new challenges to service providers in terms of service lifecycle
management and automation to address the QoS and service availability
requirements of heterogeneous applications. To this purpose, close control loops
and techniques are required towards providing automation, resource usage
optimization and reliability eventually leveraging network analytics assisted
decisions [27]. In this direction, ONAP is working on new solutions for
providing automation, performance optimization and, in general, service
lifecycle management capabilities [28]. On the other hand, the reliability of
service chains is stated as a primary requirement to assure proper service
availability [29][30]. However, the problem to address QoS and service chaining
reliability is challenging due many different and heterogeneous application

requirements. A way to effectively address reliability is to assure adaptive
resource provisioning and protection mechanisms while service chains runs
aiming at preventing service degradations due to the concurrent use of resources
from different applications [31][32]. Moreover, application-oriented
mechanisms are desirable that can be achieved through intent-based approach
[33] and detection of service degradation on end-to-end basis [32].
References
1. https://datatracker.ietf.org/meeting/99/materials/slides-99-netslicing-alex-galis-netslicing-terms-and-
systems
2. NGMN Alliance, “Description of Network Slicing Concept”, NGMN 5G P1 Requirements &
Architecture Work Stream End-to-End Architecture, v1.0.8, Sep. 14, 2016,
https://www.ngmn.org/fileadmin/user_upload/161010_NGMN_Network_Slicing_framework_v1.0.8.pdf
(accessed Apr. 30, 2018)
3. “3rd Generation Partnership Project; Technical Specification Group Services and System Aspects;
Telecommunication management; Study on management and orchestration of network slicing for next
generation network (Release 15)”, 3GPP TR 28.801 V15.1.0 (2018-01)
4. ETSI GR NFV-EVE 012, “Network Functions Virtualisation (NFV) Release 3; Evolution and
Ecosystem; Report on Network Slicing Support with ETSI NFV Architecture Framework” v3.1.1,
December 2017.
5. https://datatracker.ietf.org/wg/netslicing/about/ (accessed april 30, 2018)
6. L. Qiang et al., “Technology Independent Information Model for Network Slicing draft-qiang-coms-
netslicing-information-model-02, https://tools.ietf.org/html/draft-qiang-coms-netslicing-information-
model-02, Expiration date: July 30, 2018 (accessed April 30, 2018)
7. “Applicability of Abstraction and Control of Traffic Engineered Networks (ACTN) to Network Slicing
draft-king-teas-applicability-actn-slicing-02”, https://tools.ietf.org/html/draft-king-teas-applicability-actn-
slicing-02, Expiration date: July (accessed April 30, 2018)
8. https://www.broadband-forum.org/5g
9. ONF, “TR-526 Applying SDN Architecture to 5G Slicing”, April 2016
10. Yoshinori Goto, “Activity Report of ITU-T Focus Group on IMT-2020”, NTT Technical Review, Vol.
15 No. 6 June 2017
11. http://www.mef.net/Assets/Technical_Specifications/PDF/MEF_55.pdf (accessed April 30, 2018)
12. https://www.mef.net/Assets/White_Papers/MEF_Third_Network_LSO_Vision_FINAL.pdf (accessed
April 30, 2018)
13. https://www.gsma.com/futurenetworks/5g/introduction-to-5g-network-slicing/ (accessed April 30, 2018)
14. 5G-Transformer website, http://5g-transformer.eu/
15. 5G!Pagoda, “D2.3: Initial report on the overall system architecture definition”, https://goo.gl/2Z7PWu,
June 30, 2017
16. SONATA, “D2.2 Architecture Design”,
17. 5G PPP, “View on 5g architecture,” White Paper, https://5g-ppp.eu/white-papers (accessed 10/07/2017).
18. V. G. Nguyen et al., "SDN/NFV-Based Mobile Packet Core Network Architectures: A Survey," in IEEE
Communications Surveys & Tutorials, vol. 19, no. 3, pp. 1567-1602, third quarter 2017.
19. 3GPP TR 38.801, “Study on new radio access technology: Radio access architecture and interfaces
(Release 14)”, V14.0.0 (2017-03).
20. 3GPP TS 23.007, “Technical Specification Group Core Network and Terminals; Restoration
procedures”, Dec 2017.
21. L. Valcarenghi, et al., “Quality-of-service-aware fault tolerance for grid-enabled applications”, Optical
Switching and Networking, Volume 5, Issues 2–3, 2008.
22. ETSI GS NFV-REL 002 V1.1.1 (2015-09), “Network Functions Virtualisation (NFV); Reliability;
Report on Scalable Architectures for Reliability Management”.
23. F. F. Moghaddam, et al., "Self-Healing Redundancy for OpenStack Applications through Fault-Tolerant
Multi-Agent Task Scheduling," IEEE CloudCom 2016
24. EU SoftFIRE project, https://www.softfire.eu/
25. M. Gharbaoui, C. Contoli, G. Davoli, G. Cuffaro, B. Martini, F. Paganelli, W. Cerroni, P. Cappanera, P.
Castoldi, “Experimenting latency-aware and reliable service chaining in Next Generation Internet testbed
facility,” IEEE Conference on Network Function Virtualization and Software Defined Networks (NFV-
SDN), 2018.
26. K. Antevski, et al., “Resource Orchestration of 5G Transport Networks for Vertical Industries,” arXiv
preprint arXiv:1807.10430, 2018.
27. N. Sousa, et al., “Network Service Orchestration: A Survey, ” 2018.
28. http://github.com/onap/clamp
29. A. Hmaity, et al., "Virtual Network Function placement for resilient Service Chain provisioning," 8th
International Workshop on Resilient Networks Design and Modeling (RNDM), Halmstad, 2016.
30. M. Gharbaoui, C. Contoli, G. Davoli, G. Cuffaro, B. Martini, F. Paganelli, W. Cerroni, P. Cappanera, P.
Castoldi, “Demonstration of Latency-Aware and Self-Adaptive Service Chaining in 5G/SDN/NFV
infrastructures,” IEEE Conference on Network Function Virtualization and Software Defined Networks
(NFV-SDN), 2018.
31. AA. Mohammed, M. Gharbaoui, B. Martini, F. Paganelli, P. Castoldi, “SDN controller for network-
aware adaptive orchestration in dynamic service chaining,” IEEE NetSoft Conference and Workshops
(NetSoft), Seoul, 2016.
32. M. Gharbaoui, S. Fichera, P. Castoldi, B. Martini, “Network orchestrator for QoS-enabled service
function chaining in reliable NFV/SDN infrastructure,” ," IEEE NetSoft Conference and Workshops
(NetSoft), 2017.
5G Localization and Context-Awareness
Stefania Bartoletti, Andrea Conti, Davide Dardari, and Andrea Giorgetti
The fifth generation (5G) wireless ecosystem will be essential for a myriad of new
applications based on accurate location awareness and other contextual information.
Such wireless ecosystem will be enabled by advanced wireless technologies, e.g.
new radio (NR), integrated with existing technologies for the Internet-of-things (IoT)
and the global navigation satellite system (GNSS). First, we will explore the main
5G use cases presented by the third generation partnership project (3GPP) where
accurate positioning is required. Second, the main technologies are described. Then,
foundations and signal processing techniques for accurate localization are presented.
Finally, some context-aware applications beyond localization are discussed.
1 Localization Use Cases and KPIs
1.1 Use Cases
The 3GPP categorizes the main localization use cases based on verticals, briefly
summarized in the following.
Stefania Bartoletti
CNIT, DE, University of Ferrara, e-mail: stefania.bartoletti@unife.it
Andrea Conti
CNIT, DE, University of Ferrara, e-mail: andrea.conti@unife.it
Davide Dardari
CNIT, DEI, University of Bologna, e-mail: davide.dardari@unibo.it
Andrea Giorgetti
CNIT, DEI, University of Bologna, e-mail: andrea.giorgetti@unibo.it
167
168 Stefania Bartoletti, Andrea Conti, Davide Dardari, and Andrea Giorgetti
1.1.1 Regulatory and Mission Critical
• Accurate localization within an emergency call service refers to the geolocation

of an user when contacting a public safety answering point (PSAP) through a
short dial emergency telephone number (e.g., E-911);
• Accurate localization of a first responder who is injured or incapacitated during
mission critical operations has long been a goal of public safety. A mission critical
service enables a first responder to stay in contact with other first responders as
well as dispatch and command/control during normal and critical operations;
• The alerting service of nearby emergency responders refers to the alert of all
qualified individuals within close vicinity of the victim via their phones with a
request to provide urgent care in case of a medical emergency; The purpose of
this use case is to improve the localization of the emergency responders closest
to the victim, in order to safeguard the quickest availability of care;
• The localization service of emergency equipment outside hospitals refers to the
localization of life-saving medical equipment, such as automated external defib-
rillators (AEDs), deployed throughout public and private spaces in case of need.
This use case is about knowing where the equipment actually is, rather than where
it is supposed to be.
1.1.2 Location-Based Services
• The bike sharing service allows a rider to rent a bike via a mobile app and drop
it off anywhere for the next user. The accurate locations of shared bikes that are
available is required by the riders to find the nearest bike;
• The localization of users is a key service for augmented reality together with the
estimation of motion. Moreover, the access to databases of contextual information
and geo-localized information systems (GIS) need to be provided with low latency;
• The wearable devices such as smart watches can replace mobile terminals to
provide the customer with basic services such as tracking, activity monitoring,
and emergency messages. However, they require higher power durability in order
to replace smart terminals in applications that depend on accurate localization;
• Localization for advertisement push refers to the advertisement the relies on data
analysis of human activity location. For the advertisement to be effective, it needs
to be closely related to the user profile and location in a period of time;
• The location-based flow management refers to the use of location data of people
in public spaces or any transportation hub (airports, metro or rail station, etc.)
facing large passenger flows to elaborate statistics on passengers as well as to
optimise their organisation and signalling to passenger.
5G Localization and Context-Awareness 169
1.1.3 Industry and eHealth
• The localization of persons and medical equipment in hospitals in real time (in-
door and outdoor, even in the presence of large green areas and several buildings)
is important, for example, to notify the medical staff if patients reach a non-
authorised area and to locate caregivers and medical equipment (e.g. crash cart),
especially in emergency situations;
• The localization of patients outside hospitals refers to patients who manage to
leave the hospital without authorisation and ambulatory patients with a potentially
critical condition (e.g. cardiac, diabetes, high-risk pregnancy);
• For a smarter and more efficient waste collection and management, location data
is essential for finding optimal routes for collection vehicles and for locating bins
where sensors detected fires or other anomalies.
1.1.4 Transport (Road, Railway, Maritime, and Aerials)
• Localization for traffic monitoring, management, and control refers to vehicles and
their location on a map of the infrastructure (roads, lanes). The vehicle position
information needs to be managed over multiple road segments and long distances.
This use case addresses a more dynamic implementation, complementing sensor
and videos with position-related data determined using the 5G system;
• Road-user charging (RUC) defines generic services monitoring vehicle positions
(and/or motion) with the aim of levying a charge or a tax based on the way the
road infrastructure is used;
• The tracking of asset and freights has a key role to optimize the overall transporta-
tion efficiency, and to improve end-to-end traceability. Freight tracking enables
more accurate scheduling of all involved operations, while asset tracker should
fulfil very long lifetime (up to 15 years) and position-related data need to be
secured and protected against tampering;
• The accurate positioning of unmanned aerial vehicle (UAV) is important to sup-
port their missions and operations. Each UAV needs to be geo-localized with
high accuracy (absolute position information) to contextualize data collected in
the monitored area (e.g. images of the environment that is flown over).
In specific applications, the network operator can be asked to provide a customized
localization service with different performance for different users. Therefore, the
support of multiple different localization services can be considered as a use case
itself. This can be obtained by relying on multiple technologies for example, 3GPP
technologies and non-3GPP technologies. Different localization methods support
different levels of accuracy capabilities as described in Sec. 2. So, it is suggested
to support negotiation of localization capabilities considering user, application, or
network operator’s demands.
1.2 Key Performance Indicators and Key Attributes
The 3GPP introduces several location-based key performance indicators (KPIs) for
5G applications. The KPIs are defined based on an absolute or a relative position
estimation, which can be further specialized into an horizontal position (referring
to the position in a 2D reference or horizontal plane) and into a vertical position
(referring to the position on the vertical axis or altitude) [1]. In some applications,
the availability of position estimates is an additional attribute that describes the
percentage of time when a positioning system is able to provide the required position
data within the performance targets or requirements.
In [1], three KPIs are defined for the accuracy of parameter estimation: (1) position
accuracy describes the closeness of the estimated position of the user equipment (UE)
(either of an absolute position or of a relative position) to its true position; (2) speed
accuracy describes the closeness of the estimated magnitude of the UE’s velocity to
the true magnitude; (3) bearing accuracy describes the closeness of the measured
bearing of the UE to its true bearing. For a moving UE, the bearing is a measure of
the velocity’s direction and this KPI can be combined with speed accuracy into the
velocity’s accuracy.
Other three KPI are related to the timing of parameter estimation availability: (1)
latency describes time elapsed between the event that triggers the determination of
the position-related data and their availability at the positioning system interface; (2)
time to first fix (TFF) describes the time elapsed between the event triggering for the
first time the determination of the position-related data and their availability at the
positioning system interface; (3) update rate is the rate at which the position-related
data is generated by the localization system. It is the inverse of the time elapsed
between two successive position-related data.
Moreover, two KPIs are related to the energy consumption for localization: (1)
power consumption indicates the electrical power (usually in mW) used by the
localization system to produce the position-related data.; (2) energy per fix indicates
the electrical energy (usually in mJ per fix) used by the localization system to
produce the position-related data. It represents the integrated power consumption of
the positioning system over the required processing interval, and it considers both the
processing energy and the energy used during the idle state between two successive
productions of position-related data. This KPI can advantageously replace the power
consumption when the positioning system is not active continuously (e.g. device
tracking).
Finally, the system scalability defines the amount of devices for which the posi-
tioning system can determine the position-related data in a given time unit, and/or
for a specific update rate.
Table I summarizes the localization KPIs requirements for positioning use cases
organized per verticals.
Fig. 1 Potential requirements per use case highlighted by the 3GPP Rel. 16
2 Technologies for Cellular Network Localization
In principle, any signal propagating in a wireless environment intrinsically con-

veys position-dependent information that can be exploited for localization. Such
a position-dependent information can be extracted from measurements of signal
metrics such as received signal strength indicator (RSSI), time-of-arrival (TOA),
angle-of-arrival (AOA), phase, or combinations of them, depending on the radio
technology. One or multiple receivers compute signal measurements with respect to
one or multiple reference transmitters, and then infer the position by means of a
localization algorithm.
Satellite navigation has been considered the main technology for localization of
mobile devices so far, thanks to its global coverage and good accuracy [2]. Cellular-
based localization is used in the presence of blockage and NLOS conditions of the
satellite signals (e.g., urban and indoor environments). The positioning methods can
be classified into two main categories depending on the entity that computes the
position: (1) mobile-based: the mobile device infers its own location by measuring
signals transmitted by satellite transmitters, cellular base stations (BSs), and other
mobile nodes; (2) network-based: the network location server infers the position of
the mobile device from measurements performed by the network with respect to the
mobile device, or signal measurements performed and sent by the mobile device to
the network. Classical signal processing techniques for cellular networks include:
• Proximity: the position of the mobile terminal is approximated with the position
of the transmitter. For example, in the cell-ID method, the position of the mobile
terminal is defined as the one of the serving base station;
• Fingerprinting: the algorithm is based on finding the best match for a certain
signal metric, such as RSSI, TOA or channel delay spread, from a database of
fingerprints associated with specific locations;
• Trilateration: the position estimate is obtained by intersecting geometric forms
with parameters related to signal metrics. For example, eometric forms can be
circles or hyperbolas defined by TOA or AOA.
A combination of the previous localization algorithms can be implemented to
improve the overall performance, or to support an algorithm that cannot be computed
stand-alone given the lack of signal measurements. More advanced techniques based
on soft information rather than in single value estimates are described in [3].
The choice of the technology and the type of measurements affect the complexity
of the localization process and the KPIs.
2.1 Localization in 1G - 4.5G Cellular Networks
2.1.1 1G - 3G Cellular Networks
Positioning of UEs exploiting the pervasive diffusion of existing cellular networks,

hence without the necessity to deploy ad hoc and expensive wireless infrastructures,
has been discussed since the introduction of the first generation (1G) for vehicles
location (a comprehnsive survey can be found in [2]). In fact, at that time, UEs
were not equipped with global positioning system (GPS) receivers or, if any, they
were costly. In subsequent years, one of the main driving application was emergency
positioning (e.g., E-911 in U.S.) which became a mandatory requirement of the
U.S. Federal communications commission (FCC). In this context, cellular networks
operators are responsible for positioning UEs with an accuracy of 100 m and 300 m
in 67% and 90% of all positioning attempts, respectively.
Unfortunately, 1G-3G cellular standards were initially designed and optimized
having in mind data and voice communication services but not positioning. Never-
theless, due to the increasingly demand of location-based servicess (LBSs), in 2008
the specifications of positioning methods to be supported in global system for mobile
communications (GSM), also known as 2G, and universal mobile telecommunica-
tions system (UMTS), also known as 3G, were included in the 3GPP standardization
process [4] through the definition of some classes of location services and the in-
troduction of some approaches designed to provide position information using the
available signals structures. The main methods to estimate the position in 1G-3G
cellular networks are:
• Cell-ID: The position of a UE connected to a specific BS, which is identified
by its cell ID, is determined by the location of the BS itself (proximity). This is
a very rough method of positioning which can be improved by considering the
center of gravity of multiple BSs seen by the UE (up to 7 in GSM);
• RSSI: RSSI measurements can be used to infer the distance between the UE and
the BS. Unfortunately, propagation effects make the correlation between RSSI
and distance weak thus leading to errors in the order of 150-200 meters. RSSI
measurements can be exploited for trilateration or in fingerprinting techniques, or
radiofrequency pattern matching (RFPM), where RF maps can be created by an
advanced radio propagation prediction software, possibly refined by surveying,
and exploited through pattern matching algorithms to determine the UE position.
Fingerprinting can be in principle very accurate but it requires frequent RF maps
updates and it is very sensitive to changes in propagation conditions;
• Mobile-Assisted TOA: In GSM a rough measurement of the signal round-trip time
(timing advance (TA)) is provided in order to synchronize the UE with the BS
timing of slots. When combined with Cell-ID, such measurements can slightly
increase the positioning performance even though TA is available only during the
call at the UE;
• OTDOA: The observed time difference-of-arrival (OTDOA) is the time-difference
between the system frame numbers (SFNs) generated by two BSs as observed
Accuracy
1Km
Long-range IoT CID
3G TDOA
100m E-CID
RFPM
10m 4G TDOA
A-GNSS
WLAN/BT
1m RFID Future standards
UWB
10cm
Coverage
indoor outdoor urban outdoor rural
Fig. 2 Qualitative accuracy and coverage of positioning technologies.
by the UE. These measurements, together with other information concerning the
position of the involved BSs and the relative time difference (RTD) of the actual
transmission of the downlink signals, is used to estimate the position of the UE.
Since each OTDOA measurement related to a pair of BSs describes a line of
constant TOA difference, yielding a hyperbola in two dimensions, UE position
is determined by the intersection of hyperbolas of at least two pairs of BSs.
Clearly, OTDOA is a UE-based positioning method which requires a specific
implementation at the UE;
• Assisted - GNSS: Cellular network standard protocols have allocated resources
to carry GNSS assistance data to GNSS-enabled mobile devices in both GSM
and UMTS networks. The purpose is to assist the receiver in improving the
performance in terms of startup time, sensitivity and power consumption.
A comparison of different approaches is qualitatively illustrated in Fig. 2.
2.1.2 4G and 4.5G Cellular Networks
The first long-term evolution (LTE) Release 8 did not provide positioning protocols.
3GPP boosted location services in LTE Release 9, delivered in December 2009 [4],
with particular emphasis to emergency calls, as required by FCC E-911. Positioning
methods in LTE networks can be dependent on the radio access technique (RAT),
that is making use of LTE signals, or independent of the RAT, that is using other
signals such as GPS.
As can be seen in Fig. 3, most of RAT-dependent positioning methods are similar
to those used in UMTS [5]. E-CID is an improved version of Cell-ID in which
Positioning technology
Cell-ID X X
RAT-dependent
E-CID X X X X X
OTDOA X X X X X X
UTDOA X X X X
RFPM X X
A-GNSS X X X X X X X
RAT-independent
TBS X
WLAN X
Bluetooth X
Barometer X
2G 3G 3.9G 4G 4.5G 5G
LTE-A pro (R12)
LTE-A pro (R13)

LTE-A (R11)
LTE-A (R9)
LTE (R8)
GPS
UMTS
Fig. 3 Positioning methods in cellular network standards.
cell ID information is combined with other measurements such as TA, round-trip

time, and AOA. In LTE, OTDOA uses specific downlink signals called positioning
reference signals (PRSs) which are transmitted in certain positioning subframes
of the orthogonal frequency-division multiple access (OFDMA) signal structure
grouped into positioning occasions which occur periodically every 160, 320, 640
or 1280 ms. PRS are received by the UE so that it can perform TOA measurements
(see Fig. 4). The UE measurement is known as the reference signal time difference
measurement (RSTD) which represents the relative time difference between two
BSs. The UE reports its RSTD measurements back to the network, specifically to
the location server, which determines the position of the UE.
In LTE Release 11, the uplink time difference-of-arrival (UTDOA) has been
introduced, thus allowing the network of BSs, also known as eNB in LTE, to collect
time difference-of-arrival (TDOA) measurements of the signal transmitted by the UE
and hence localize it. The UTDOA method is based on network measurements of the
TOA, in at least 3 BSs, of the signal transmitted by the UE. The difference between
two TOAs at two BSs defines a hyperbola and the position of the UE can be calculated
as in the OTDOA method. The main difference is that now all the processing is done
by the network and no new functionalities need to be implemented in the UE.
FCC recognized that positioning requirements for indoor scenarios cannot be met
by most of operators, thus new requirements were released in 2015. Specifically,
BS 1
OTDOA BS1-BS3
hyperbola
PRS
BS 3
PRS
BS 2
PRS
RSTDs
Location RSTDs OTDOA BS2-BS3

Server hyperbola
Core network
Fig. 4 OTDOA-based positioning in LTE.
a 50 m horizontal accuracy should be provided for 40, 50, 70, and 80% of emer-
gency calls within 2, 3, 5, and 6 years respectively. For vertical performance, the
operators should propose an accuracy metric within 3 years. In response, most of
RAT-independent positioning methods have been specified in LTE Release 13 (LTE-
Advanced pro, 4.5G) with the purpose to enhance the positioning accuracy, especially
in indoor environments, as required by FCC rules. This was made possible using mul-
tiple different technologies such as wireless local area network (WLAN)/Bluetooth,
barometric pressure sensors (vertical positioning), and terrestrial beacon systems
(PRS beacons and metropolitan beacon systems). Also RAT-dependent methods,
in particular OTDOA, have been enhanced by defining new PRS patterns and PRS
bandwidth extension.
2.2 Localization in 5G Cellular Networks
The main difference between 5G and previous standards is that 5G KPIs requirements
are no longer defined by the regulatory body for emergency calls, but they are driven
by the 5G use-cases as described in Sec. 1 and are being used in standardization [6].
The KPIs for accuracy, latency, and energy consumption are reported in Sec. 1.2.
The standardization of positioning in 5G is still under discussion within dedicated
task in Release 16 [1]. Localization will be based on the characteristics of the up-link
Massive antenna array
Fig. 5 Single-anchor positioning with massive mmWave antenna arrays.
and down-link signals of NR (3GPP-technologies) but also on new technologies and

network configurations, for example, GNSS (e.g. BeiDou, Galileo, GLONASS, and
GPS), Terrestrial Beacon Systems (TBS), Bluetooth, WLAN, RFID, and sensors [7].
The main breakthrough in 5G is due to the employment of massive multiple-
input–multiple-output (MIMO) beamforming and of millimeter wave (mmWave)
signals. The use of mmWave brings a two-fold advantage: large available bandwidth
and the possibility to pack a large number of antenna elements even in small spaces
(e.g., in a smartphone). Wideband signals offer better time resolution and robustness
to multipath thus improving the performance of OTDOA/UTDOA schemes, as well
as paving the way to new positioning methods such as multipath-assisted localization
exploiting specular multipath components to obtain additional position information
from radio signals [8]. A large number of antenna elements enables massive MIMO
and very accurate beamforming (see Fig. 5). This will make possible the introduction
of single-anchor approaches providing cm-level and degree-level accuracy in 6D
positioning (3D position and 3D orientation) [9], thus overcoming the problem of
deploying a redundant ad-hoc infrastructure which is, nowadays, a major bottleneck
for the widespread adoption of indoor localization systems. In addition, device-to-
device (D2D) are under consideration in Release 16 for ultra-dense networks enabling
cooperative localization, for instance, in vehicle-to-everything (V2X) scenarios [10].
2.3 Non-3GPP Technologies for 5G Localization
The lack of service coverage of GNSSs in indoor environments has generated a rich
research activity on the design of indoor localization solutions in the last two decades.
Some solutions exploit acoustic, infrared, laser, inertial, and vision technologies,
whereas others are based on measurements of specific features of radio signals (e.g.,
TOA, RSSI, etc.) [11].
In the context of radio-based positioning technologies, research efforts followed
two main directions: exploitation of existing standards designed only for communi-
cation; and design of ad-hoc standards/solutions for positioning. Recently, particular
emphasis has been given to technologies for IoT applications which typically use
low-cost, low-complexity, and low-energy devices.
2.3.1 Communication-designed Technologies
Several wireless technologies and standards are currently available for WLANs,
wireless sensor networks (WSNs) and IoT applications in general. Examples are Wi-
Fi, radiofrequency identification (RFID), ZigBee and Bluetooth low energy (BLE).
They do not offer specific positioning capabilities, but their transmitted signals can
be exploited to provide different localization performance levels. While RFID and
BLE, due to their limited range, are typically used with proximity methods, Wi-Fi
technology has been successfully adopted in several positioning systems typically
leveraging on fingerprinting methods where meter-level accuracies can be achieved in
many conditions. Wi-Fi ands BLE have already been considered as complementary
technologies in LTE Release 13 to enhance positioning in indoor environments,
especially thanks their wide diffusion.
2.3.2 Ad-hoc Technologies
The most promising ad-hoc technology for high-accuracy positioning in indoor

environment is ultra-wideband (UWB) [12]. This is justified by the fact that the
larger is the signal bandwidth the higher will be the resolution of time measurements
and hence positioning [13, 14].
According to the FCC, an UWB signal is defined as a signal that has fractional
(relative) bandwidth larger than 20% or an absolute bandwidth of at least 500 MHz
[15]. UWB signals are efficiently generated using impulse generators that are simple
and energy efficient. After the FCC allowed the use of UWB signals in 2002 in the
U.S. and the same happened in Europe in 2007 [16] and worldwide, standardization
efforts took place. This effort resulted in the publication of the IEEE 802.15.4a
standard in 2007 [17] as a physical layer (PHY) alternative based on UWB to the
IEEE 802.15.4 standard for WSNs (the PHY of ZigBee). This standard was followed
by the IEEE 802.15.4f standard published in 2012 for applications in the field of port
& marine cargo, automotive, logistics, industrial and manufacturing [18] .
After an initial slow market penetration, mainly caused by the high-cost of pro-
prietary devices and the fragmented worldwide power emission mask regulations,
since 2014 the market of real time locating systems (RTLS) took off, thanks to the
availability of low-cost chips compliant with the IEEE 802.15.4a standard, and its
growing rate is around 40% yearly, especially in the field of logistic and Industry
4.0. Recently, UWB has been coupled with the RFID technology to detect and track
battery-less tags powered via wireless links [19]. Besides active positioning, thanks
to its peculiarities, the UWB technology enables also other applications like multi-
static radar for non-collaborative localization [20], life signs detection systems, and
through-wall and underground imaging as will be discussed in Sec. 2.3.4.
2.3.3 Long-range IoT Applications
Most of long-range IoT applications (e.g., smart city, asset tracking, smart metering,
smart farming, and smart logistics) are low-rate applications with coverage of tens of
kilometers, and require battery life lasting years (in some cases more than 10 years).
Moreover, nodes have small capabilities in terms of computation and memory, which
makes accurate localization challenging [21]. Currently, two proprietary solutions
are emerging: LoRa and Sigfox. Both have very low throughputs from few tens of
bits per second up to few hundreds of kbps. They are not designed for positioning and
employ narrowband signals that make time measurements very inaccurate because
of the consequent scarce temporal resolution. Despite that, recent studies showed
that rough positioning accuracy in the order of hundred of meters are possible by
properly processing TDOA measurements at BSs [22].
The IoT market is under consideration also by the standardization bodies. The
two main standard technologies for long-range IoT solutions are IEEE 802.11 Long
Range Low Power (LRLP) and the 3GPP narrowband technologies, i.e., LTE-M,
LTE NB-IoT, and EC-GSM-IoT. The positioning capabilities of 3GPP narrowband
technologies were investigated in LTE Release 14. The main positioning algorithms
are enhanced-CID (ECID), OTDOA and UTDOA with a target accuracy of 50 m [23].
The 5G standard is expected to include dedicated protocols for positioning to enable
positioning in IoT applications, even though several technical issues have still to be
studied [24].
2.3.4 Integration with Device-free Localization
An increasing attention is recently being devoted to the capability of detecting and

tracking objects that do not take actions to help the localization infrastructure or that
do not wish to be detected and localized at all. This operation is referred to non-
collaborative localization and has often been undertaken by exploiting a network
of radio sensors able to scan the area of interest through wideband radio signals
to create a radio image of the objects and the environment. These systems are
classified based on whether the network emits a signal designed for target detection
In-sensor processing
− Demodulation
− ToA estimation
− Clutter removal
− Ghost mitigation
− CFAR detection
− 1D clustering
− Measurement select.
y Objects trajectory
Network processing
− Detection & localization

− 2D measurement selection
− 2D clustering
− Object tracking
− Target association
1
Fig. 6 A scenario for object/people tracking through a sensor radar network. The processing steps
are performed both locally on sensors and at network level.
and localization (active radar) [25], or the network exploits signals emitted by other
sources of opportunity (passive radar) [26].
Accurate localization via sensor radars becomes particularly challenging in indoor
environments characterized by dense multipath, clutter, signal obstructions (e.g., due
to the presence of walls), and interference. In a real-world scenario measurements
are usually heavily affected by such impairments, severely affecting detection re-
liability and localization accuracy. These operating conditions may be mitigated
by the adoption of waveforms characterized by large bandwidth, e.g., UWB ones
(see Sec. 2.3.2), exploiting prior knowledge of the environment, selecting reliable
measurements, and using various signal processing techniques [27–30]. The UWB
technology, and in particular its impulse radio version characterized by the transmis-
sion of a few nanoseconds duration pulses [31], offers an extraordinary resolution
and localization precision in harsh environments, due to its ability to resolve mul-
tipath and penetrate obstacles. These features, together with the property of being
light-weight, cost-effective, and characterized by low power emissions, have con-
tributed to make UWB an ideal candidate for non-collaborative object detection
in short-range radar sensor networks applications. A sketch of a scenario with the
localization of objects by a radar sensor network is depicted in Figure 6.
Ubiquitous deployment of sensor radar systems integrated with existing commu-
nication infrastructure is expected to open new application scenarios, some of which
have much in common with the use cases of Table 1. For example, through wall
imaging, i.e., the ability to locate indoor moving targets with sensors at a standoff
range outside buildings [32, 33], search and rescue of trapped victims, and people
counting [34,35] are just a few examples of promising non-collaborative localization

applications.
3 Localization Design in 5G and Beyond
The KPI requirements in 5G scenarios can be fulfilled by exploiting different signal

processing techniques and technologies, which mostly have limited resources for
communication, processing, and memory. The reliability of 5G localization lies in
fusing data and measurements collected from heterogeneous sensors with contextual
information [36] and in designing efficient network operation strategies [37].
3.1 From Foundation to Operation
3.1.1 Fundamental Limits
To provide performance benchmarks and to guide efficient network design and op-
eration, it is important to understand the fundamental limits of localization accuracy
in 5G as well as the corresponding approaches to achieve such accuracy. For this
purpose, the information inequality can be applied to determine a lower bound for the
estimation errors, which is known as the Cramér-Rao lower bound (CRLB), through
the inverse of the Fisher information matrix (FIM) [12].
To evaluate the localization performance in the presence of noise, CRLB-type
performance bounds for the signal metrics under test, e.g., TOA, OTDOA, UTDOA,
RSSI, or AOA are usually considered. Nevertheless, the properties of the signal
metrics depends heavily on the method used to infer user positions, and the use
of certain signal metrics may discard relevant information for localization. Thus, in
deriving the fundamental limits of localization accuracy, it is desirable to fully exploit
the information contained in the received waveforms rather than using specific signal
metrics extracted from the waveforms [12].
Given the complexity of the scenarios considered, the analysis of fundamen-
tal limits for 5G localization should take into account also for multipath and non
line-of-sight (NLOS) propagations which impact localization accuracy especially in
harsh propagation environments (e.g., indoor) [38]. In addition, the case of D2D
cooperation where intranode measurements are available can be analyzed by taking
into account spatial cooperation (together with temporal cooperation in dynamic
scenarios) by characterizing the information evolution in both spatial and temporal
domains [39].
3.1.2 Network Operation Strategies
The performance of localization in 5G depends on the transmitting energy, signal

bandwidth, network geometry, and the channel conditions [40, 41]. Such factors
are driven by the network operation strategy, which determines the allocation of
resources, the network nodes from which the measurements are taken, and the
deployment of mobile nodes and base stations. Network operation plays a critical role
in localization since it not only affects the energy consumption, which is a crucial KPI
in 5G applications, but also determines the localization accuracy [37]. For example,
range measurements between two nodes with poor channel conditions consume
significant amounts of energy without improving localization accuracy [28].
Network operation strategies for efficient localization and navigation can be cate-
gorized into several functionalities, including node prioritization (i.e. prioritization
strategies for allocating transmitting resources such as power, bandwidth, and time to
achieve the best trade-off between resource consumption and localization accuracy),
node activation (i.e. activation strategies for determining the nodes that are allowed
to make inter-node measurements so that the localization accuracy of the entire net-
work is maximized), and node deployment (deployment strategies for determining
the positions of new nodes in the network so that the localization accuracy of certain
existing nodes can be maximally improved) [21].
3.2 Advances in Signal Processing
The integration of hybrid technologies fusing measurements from different sensors

such as inertial, GNSS, camera, various RATs, is a challenge for signal processing
in 5G localization. In particular, new localization methods, for instance based on
statistical machine learning, will be needed to achieve accurate, seamless, and robust
localization.
3.2.1 Soft Information
Conventional localization methods rely on single-value metrics such as TOA, RSSI,

RSSI, or AOA, as described in Sec. 2. Localization accuracy depends heavily on
the quality of the single value estimates (SVEs), which degrades in the presence
of multipath and NLOS. Therefore, a big research effort focused on improving the
estimation of single value metrics [42–45]. For example, techniques to refine the
SVE or mitigate the SVE errors have been developed by relying on error models
(e.g., the bias induced by NLOS conditions) [43, 46]. In addition, features extracted
from the received waveforms can be used to selecting a subset of the waveforms that
contain reliable positional information and therefore mitigate SVE errors [28]. Also,
the SVE of different features can be fused together, for example by involving hybrid
models that account for the relationship among them [47–50].
More recently, one-stage techniques have been explored to overcome the limi-
tations of SVE. One-stage techniques obtain position estimates from the received
waveforms based on a prior model, namely direct positioning direct positioning (DP),
without requiring intermediate estimations of signal metrics [51–56]. For example,
a set of possible values rather than on single distance estimate (DE), referred to as
soft range information (SRI), can be used for localization. In general, localization
performance can be improved by designing localization networks thatexploit soft
information (SI), such as SRI or soft angle information (SAI), together with environ-
mental information, such as contextual data including digital map, dynamic model,
and users profiles [3].
The 5G and IoT scenarios offer the possibility to exploit different sensors in the
environments. The reliability of multi-sensor IoT lies in fusing data and measure-
ments collected from heterogeneous sensors with stringent limitations in terms of
energy and power consumption [21], and in designing efficient network operation
strategies [37]. This calls for distributed implementation of SI-based localization,
which require the communication of messages with high dimensionality depending
on the kind of SI [57–59]. Therefore, it will be of utmost importance to develop
dimensionality reduction techniques for efficient network localization in 5G and IoT
scenarios.
3.2.2 Cooperative localization with D2D communication
D2D communication in 5G are under consideration in Release 16, in particular

for ultra-dense networks enabling cooperative localization, for instance, in V2X
scenarios [10]. Joint spatial and temporal cooperation of devices can yield dramatic
localization performance improvement over conventional approaches since intranode
measurements and mobility (dynamic) models yield new information for localization
and navigation. In particular, joint spatial and temporal cooperation between devices
incurs in associated costs such as additional communication and more complicated
algorithms over the network: 1) the communication among nodes is required for
inter-node measurements and information exchange; and 2) interdependency among
the estimates of the agent positions hinders effective distributed inference algorithms.
In [39], the concept of network localization and navigation (NLN) has been put forth
to exploit spatiotemporal cooperation among nodes. Cooperative algorithms have
been developed based on graphical models, a branch of statistics that makes infer-
ence possible over highly interrelated random variables. Despite the technological
advances in the field, there are still several issues that need to be addressed to realize
accurate, reliable, and efficient NLN.
3.3 Beyond Localization
In addition to specific use cases as flow management, traffic monitoring, advertise-

ment push (see Sec. 1), many other applications such as crowd control and detection
of emergency events are important for smart building and public protection. All these
applications rely on ready-to-use analytics that can be defined based on localization
data. The definition of such analytics will primarily leverage basic spatiotemporal
features such as crowd size and people flow. To this aim, both individual and crowd-
centric approaches can be adopted: (i) individual-centric approaches associate the
measured data to single targets/terminals, and run knowledge discovery separately
on each of them; (ii) crowd-centric approaches associate the measured data directly
to a group of users, and run a crowd-level analysis, resulting in lower dimensionality
and complexity [35]. Therefore new methods based on a crowd-centric approach
will be conceived to provide ready-to-use analytics in 5G scenarios and fulfil the
requirements in terms of latency and energy consumption.
References
1. 3GPP, “3GPP TR-2 2.872 : Study on positioning use cases,” in ETSI, Tech. Report 16, 2018.
[Online]. Available: http://www.3gpp.org/
2. J. A. del Peral-Rosado, R. Raulefs, J. A. López-Salcedo, and G. Seco-Granados, “Survey of
cellular mobile radio localization methods: From 1G to 5G,” IEEE Communications Surveys
Tutorials, vol. 20, no. 2, pp. 1124–1148, Secondquarter 2018.
3. S. Mazuelas, A. Conti, J. C. Allen, and M. Z. Win, “Soft range information for network
localization,” IEEE Trans. Signal Process., vol. 66, no. 12, pp. 3155–3168, Jun. 2018.
4. “The 3rd generation partnership project (3gpp).” [Online]. Available: http://www.3gpp.org/
5. S. S. Cherian and A. N. Rudrapatna, “Lte location technologies and delivery solutions,” Bell
Labs Technical Journal, vol. 18, no. 2, pp. 175–194, Sept 2013.
6. 3GPP, “3GPP RP-172795: Handling new SI/WI proposals in RAN,” in ETSI, Tech. Report
12, 2017. [Online]. Available: http://www.3gpp.org/
7. Y. Liu, X. Shi, S. He, and Z. Shi, “Prospective positioning architecture and technologies in
5G networks,” IEEE Network, vol. 31, no. 6, pp. 115–121, November 2017.
8. K. Witrisal, P. Meissner, E. Leitinger, Y. Shen, C. Gustafson, F. Tufvesson, K. Haneda,
D. Dardari, A. F. Molisch, A. Conti, and M. Z. Win, “High-accuracy localization for assisted
living: 5G systems will turn multipath channels from foe to friend,” IEEE Signal Processing
Magazine, vol. 33, no. 2, pp. 59–70, March 2016.
9. A. Guerra, F. Guidi, and D. Dardari, “Single-anchor localization and orientation performance
limits using massive arrays: MIMO vs.beamforming,” IEEE Transactions on Wireless Com-
munications, vol. 17, no. 8, pp. 5241–5255, Aug 2018.
10. H. Wymeersch, G. Seco-Granados, G. Destino, D. Dardari, and F. Tufvesson, “5G mmwave
positioning for vehicular networks,” IEEE Wireless Communications, vol. 24, no. 6, pp. 80–86,
Dec 2017.
11. D. Dardari, E. Falletti, and M. Luise, Satellite and Terrestrial Radio Positioning Techniques -
A signal processing perspective, 1st ed. London, UK: Elsevier Ltd, 2011.
12. Y. Shen and M. Z. Win, “Fundamental limits of wideband localization – Part I: A general
framework,” IEEE Trans. Inf. Theory, vol. 56, no. 10, pp. 4956–4980, Oct. 2010.
13. D. Dardari, A. Conti, U. J. Ferner, A. Giorgetti, and M. Z. Win, “Ranging with ultrawide
bandwidth signals in multipath environments,” Proc. IEEE, vol. 97, no. 2, pp. 404–426, Feb.
2009, special issue on Ultra -Wide Bandwidth (UWB) Technology & Emerging Applications.
14. S. Bartoletti, W. Dai, A. Conti, and M. Z. Win, “A mathematical model for wideband ranging,”
IEEE J. Sel. Topics Signal Process., vol. 9, no. 2, pp. 216–228, Mar. 2015.
15. First Report and Order 02-48, Federal Communications Commission, Feb. 2002.
16. The Commission of the European Communities, “Commission Decision of 21
February 2007 on allowing the use of the radio spectrum for equipment using
ultra-wideband technology in a harmonised manner in the Community,” Official
Journal of the European Union, 2007/131/EC, Feb. 23, 2007. [Online]. Available:
http://eur-lex.europa.eu/LexUriServ/site/en/oj/2007/l_055/l_05520070223en00330036.pdf
17. “IEEE standard for information technology - telecommunications and information exchange
between systems - local and metropolitan area networks - specific requirement part 15.4:
Wireless medium access control (MAC) and physical layer (PHY) specifications for low-rate
wireless personal area networks (WPANs),” IEEE Std 802.15.4a-2007 (Amendment to IEEE
Std 802.15.4-2006), pp. 1–203, 2007.
18. IEEE Standard for Information Technology - Telecommunications and Information Exchange
Between Systems - Local and Metropolitan Area Networks - Specific Requirements - Part 15.4:
Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Low
Rate Wireless Personal Area Networks (WPANs) - Amendment: Active RFID System PHY,
IEEE Std 802.15.4f-2012, 2012.
19. A. Costanzo, D. Dardari, J. Aleksandravicius, N. Decarli, M. D. Prete, D. Fabbri, M. Fantuzzi,
A. Guerra, D. Masotti, M. Pizzotti, and A. Romani, “Energy autonomous UWB localization,”
IEEE Journal of Radio Frequency Identification, vol. 1, no. 3, pp. 228–244, Sept 2017.
20. B. Sobhani, E. Paolini, A. Giorgetti, M. Mazzotti, and M. Chiani, “Target tracking for UWB
multistatic radar sensor networks,” IEEE J. Sel. Topics Signal Process., vol. 8, no. 1, pp.
125–136, Feb. 2014.
21. M. Z. Win, F. Meyer, Z. Liu, W. Dai, S. Bartoletti, and A. Conti, “Efficient multi-sensor
localization for the Internet-of-Things,” IEEE Signal Process. Mag., vol. 35, no. 5, pp. 153–
167, Sep. 2018.
22. B. C. Fargas and M. N. Petersen, “GPS-free geolocation using LoRa in low-power WANs,” in
2017 Global Internet of Things Summit (GIoTS), June 2017, pp. 1–6.
23. J. A. del Peral-Rosado, J. A. LÛpez-Salcedo, and G. Seco-Granados, “Impact of frequency-
hopping NB-IoT positioning in 4G and future 5G networks,” in 2017 IEEE International
Conference on Communications Workshops (ICC Workshops), May 2017, pp. 815–820.
24. X. Lin, J. Bergman, F. Gunnarsson, O. Liberg, S. M. Razavi, H. S. Razaghi, H. Rydn, and
Y. Sui, “Positioning for the internet of things: A 3GPP perspective,” IEEE Communications
Magazine, vol. 55, no. 12, pp. 179–185, Dec 2017.
25. M. I. Skolnik, Radar Handbook, 2nd ed. McGraw-Hill Professional, Jan. 1990.
26. F. Colone, D. Pastina, P. Falcone, and P. Lombardo, “WiFi-based passive ISAR for high-
resolution cross-range profiling of moving targets,” IEEE Trans. Geosci. Remote Sens., vol. 52,
no. 6, pp. 3486–3501, Jun. 2014.
27. S. Bartoletti, A. Conti, A. Giorgetti, and M. Z. Win, “Sensor radar networks for indoor
tracking,” IEEE Wireless Commun. Lett., vol. 3, no. 2, pp. 157–160, Apr. 2014.
28. S. Bartoletti, A. Giorgetti, M. Z. Win, and A. Conti, “Blind selection of representative obser-
vations for sensor radar networks,” IEEE Trans. Veh. Technol., vol. 64, no. 4, pp. 1388–1400,
Apr. 2015.
29. S. Savazzi, S. Sigg, M. Nicoli, V. Rampa, S. Kianoush, and U. Spagnolini, “Device-free radio
vision for assisted living: Leveraging wireless channel quality information for human sensing,”
IEEE Signal Process. Mag., vol. 33, no. 2, pp. 45–58, Mar. 2016.
30. M. Chiani, A. Giorgetti, and E. Paolini, “Sensor radar for object tracking,” Proc. IEEE, vol.
106, no. 6, pp. 1022–1041, 2018, (Special Issue on Foundations and Trends in Localization
Technologies).
31. M. Z. Win and R. A. Scholtz, “Impulse radio: How it works,” IEEE Commun. Lett., vol. 2,
no. 2, pp. 36–38, Feb. 1998.
32. A. Yarovoy, L. Ligthart, J. Matuzas, and B. Levitas, “UWB radar for human being detection,”
IEEE Aerosp. Electron. Syst. Mag., vol. 21, no. 3, pp. 10–14, Apr. 2006.
33. M. G. Amin, Through-the-wall radar imaging. CRC press, 2016.

34. J. W. Choi, S. H. Cho, Y. S. Kim, N. J. Kim, S. S. Kwon, and J. S. Shim, “A counting sensor for
inbound and outbound people using IR-UWB radar sensors,” in IEEE Sensors Applications
Symp. (SAS), Catania, Italy, Apr. 2016, pp. 1–5.
35. S. Bartoletti, A. Conti, and M. Z. Win, “Device-free counting via wideband signals,” IEEE J.
Sel. Areas Commun., vol. 35, no. 5, pp. 1163–1174, May 2017.
36. A. Conti, S. Mazuelas, S. Bartoletti, M. Z. Win, and W. C. Lindsey, “Soft information for
localization of things,” Proc. IEEE, vol. 107, 2019.
37. M. Z. Win, W. Dai, Y. Shen, G. Chrisikos, and H. V. Poor, “Network operation strategies for
efficient localization and navigation,” Proc. IEEE, vol. 106, no. 7, pp. 1224–1254, Jul. 2018,
special issue on Foundations and Trends in Localization Technologies.
38. K. Witrisal, P. Meissner, E. Leitinger, Y. Shen, C. Gustafson, F. Tufvesson, K. Haneda,
D. Dardari, A. F. Molisch, A. Conti, and M. Z. Win, “High-accuracy localization for assisted
living,” IEEE Signal Process. Mag., vol. 33, no. 2, pp. 59–70, Mar. 2016.
39. M. Z. Win, A. Conti, S. Mazuelas, Y. Shen, W. M. Gifford, D. Dardari, and M. Chiani,
“Network localization and navigation via cooperation,” IEEE Commun. Mag., vol. 49, no. 5,
pp. 56–62, May 2011.
40. A. Conti, M. Guerra, D. Dardari, N. Decarli, and M. Z. Win, “Network experimentation for
cooperative localization,” IEEE J. Sel. Areas Commun., vol. 30, no. 2, pp. 467–475, Feb. 2012.
41. A. Conti, D. Dardari, M. Guerra, L. Mucchi, and M. Z. Win, “Experimental characterization
of diversity navigation,” IEEE Syst. J., vol. 8, no. 1, pp. 115–124, Mar. 2014.
42. I. Güvenç, C.-C. Chong, F. Watanabe, and H. Inamura, “NLOS identification and weighted
least-squares localization for UWB systems using multipath channel statistics,” EURASIP J.
Adv. in Signal Process., vol. 2008, pp. 1–14, Article ID 271 984, 2008.
43. S. Maranò, W. M. Gifford, H. Wymeersch, and M. Z. Win, “NLOS identification and mitigation
for localization based on UWB experimental data,” IEEE J. Sel. Areas Commun., vol. 28, no. 7,
pp. 1026–1035, Sep. 2010.
44. S. Mazuelas, F. A. Lago, J. Blas, A. Bahillo, P. Fernandez, R. Lorenzo, and E. Abril, “Prior
NLOS measurements correction for positioning in cellular wireless networks,” IEEE Trans.
Vehicular Technology, vol. 58, no. 5, pp. 2585–2591, Jun 2009.
45. J. Khodjaev, Y. Park, and A. S. Malik, “Survey on NLOS identification and error mitiga-
tion problems in UWB-based positioning algorithms for dense environments,” in Annals of
Telecommunications, 2009.
46. C. Morelli, M. Nicoli, V. Rampa, and U. Spagnolini, “Hidden Markov models for radio
localization in mixed LOS/NLOS conditions,” IEEE Trans. Signal Process., vol. 55, no. 4, pp.
1525–1542, Apr 2007.
47. J. Prieto, S. Mazuelas, A. Bahillo, P. Fernandez, R. M. Lorenzo, and E. J. Abril, “Adaptive
data fusion for wireless localization in harsh environments,” IEEE Trans. Signal Process.,
vol. 60, no. 4, pp. 1585–1596, Apr. 2012.
48. A. Catovic and Z. Sahinoglu, “The Cramer-Rao bounds of hybrid TOA/RSS and TDOA/RSS
location estimaion schemes,” IEEE Commun. Lett., vol. 8, no. 10, pp. 626–628, Oct. 2004.
49. J.-C. Chen, P. Ting, C.-S. Maa, and J.-T. Chen, “Wireless geolocation with TOA/AOA mea-
surements using factor graph and sum-product algorithm,” in Proc. IEEE Semiannual Veh.
Technol. Conf., vol. 5, Sep. 2004, pp. 3526–3529.
50. P. Deng and P. Fan, “An AOA assisted TOA positioning system,” in Proc. of Int. Conf. on
Commun. Technol., vol. 2, Beijing, China, 2000, pp. 1501–1504.
51. P. Closas, C. Fernández-Prades, and J. A. Fernández-Rubio, “Direct position estimation ap-
proach outperforms conventional two-steps positioning,” in Proc. of European Signal Pro-
cessing Conference (EUSIPCO), Aug. 2009, pp. 1958–1962.
52. P. Closas and A. Gusi-Amigo, “Direct position estimation of GNSS receivers: Analyzing main
results, architectures, enhancements, and challenges,” IEEE Signal Process. Mag., vol. 34,
no. 5, pp. 72–84, Sep. 2017.
53. L. Tzafri and A. J. Weiss, “High-resolution direct position determination using MVDR,” IEEE
Trans. Wireless Commun., vol. 15, no. 9, pp. 6449–6461, Sep. 2016.
54. N. Vankayalapati, S. Kay, and Q. Ding, “TDOA based direct positioning maximum likelihood
estimator and the Cramér-Rao bound,” IEEE Trans. Aerosp. Electron. Syst., vol. 50, no. 3, pp.
1616–1635, Jul. 2014.
55. J. P. Beaudeau, M. F. Bugallo, and P. M. Djurić, “RSSI-based multi-target tracking by coop-
erative agents using fusion of cross-target information,” IEEE Trans. Signal Process., vol. 63,
no. 19, pp. 5033–5044, Oct. 2015.
56. M. Rydström, L. Reggiani, E. G. Ström, and A. Svensson, “Adapting the ranging algorithm
to the positioning technique in UWB sensor networks,” Wireless Personal Communications,
vol. 47, no. 1, pp. 27–38, Oct. 2008.
57. A. T. Ihler, J. W. Fisher III, R. L. Moses, and A. S. Willsky, “Nonparametric belief propagation
for self-localization of sensor networks,” IEEE J. Sel. Areas Commun., vol. 23, no. 4, pp. 809–
819, Apr. 2005.
58. H. Wymeersch, J. Lien, and M. Z. Win, “Cooperative localization in wireless networks,” Proc.
IEEE, vol. 97, no. 2, pp. 427–450, Feb. 2009, special issue on Ultra -Wide Bandwidth (UWB)
Technology & Emerging Applications.
59. S. Mazuelas, Y. Shen, and M. Z. Win, “Belief condensation filtering,” IEEE Trans. Signal
Process., vol. 61, no. 18, pp. 4403–4415, Sep. 2013.
IoT Support
Roberto Verdone, Silvia Mignardi
Abstract This section focuses on the Internet of Things (IoT) context in the 5G
framework. First, it introduces the IoT field with a general and incisive definition
to clarify its position in the market. Then, it describes the available technologies in
details to conclude by spanning many of the increasing potential applications.
1 The 5G Ecosystem for the IoT
When Christianity was "standardized" through the Nicea Council in 325 A.D., all
previously defined heathen festivities were absorbed (i.e. same dates used for the
"new" celebrations) by the Christian calendar, to ensure its successful assimilation
by people. 5G is doing the same, with Internet of Things (IoT) technologies.
The 5G mobile radio network is an ecosystem, made of many interdependent
elements: RATs (Radio Access Technologies), Core, Cloud, End Users, their User
Equipment (UE), Mobile Network Operators (NMOs), equipment manufacturers,
Service Providers, and other. All these elements are currently evolving while 5G
is being specified, so that, as in the past, the Fifth Generation of Mobile Radio
Networks is a step forward in an evolutionary scenario stemming from the previous
4G. However, one of the key aspects of 5G that come as a profound change with
respect to the past, is that the 5G ecosystem is specifically designed to support (also)
the IoT evolution.
The Internet of Things is a paradigm that is well known since many years. A
formal definition of the IoT can be found in a White Paper of the IEEE Internet
Initiative published in 2005 [1]: "Internet of Things envisions a self-configuring,
adaptive, complex network that interconnects Things to the Internet through the
Roberto Verdone
Università di Bologna, Viale Risorgimento 2, Bologna, e-mail: roberto.verdone@unibo.it
Silvia Mignardi
Università di Bologna, Viale Risorgimento 2, Bologna e-mail: silvia.mignardi@unibo.it
189
190 Roberto Verdone, Silvia Mignardi
use of standard communication protocols. The interconnected Things have physical

or virtual representation in the digital world, sensing/actuation capability, a pro-
grammability feature and are uniquely identifiable. The Things offer services, with
or without human intervention. The service is made available anywhere, anytime,
and for anything taking security into consideration.". The IoT has been conceived
and deployed in the past years using communication technologies defined outside
the domain of 3GPP: IEEE 802.15.4/Zigbee, Bluetooth, Sigfox, LoRa and others.
At the same time, the evolution of 2G, 3G and 4G air interfaces was more oriented
towards human-centric applications (with the exception of the recent developments
of LTE-M and NB-IoT). This paved the way to the success of the above mentioned
non-3GPP communications standards or proprietary solutions for the IoT.
The scenario has now changed. The attention that 3GPP and the main stakeholders
of the 5G ecosystem are offering to the IoT, has become huge in the recent years. The
5G network is expected to be "IoT dominated" in terms of number of devices. More
interestingly, some stakeholders are starting presenting 5G as the interface between
the physical and the digital world, thus emphasizing even more its role as enabler of
the IoT. One of the documents delivered by 3GPP within Release 16 in June 2018
[2] starts with a clear mention to the IoT: "5G: the need to support different kinds of
UEs (e.g., for the Internet of Things (IoT)), services, and technologies is driving the
technology revolution to a high-performance and highly efficient 3GPP system.".
This trend has made more and more clear that 5G has to be considered as a system
that will comprise a number of communication technologies, starting from the 5G
New Radio (NR) under development at 3GPP, including 3GPP standards like LTE-M
and NB-IOT, and the successful non-3GPP solutions that have grown in terms of
market shares in the past few years (like e.g. LoRaWAN); according to [2], "... the 5G
system shall enable the UE to select, manage, and efficiently provision services over
the 3GPP or non-3GPP access.". Like the Christianity did in 325 A.D., Christmas
will come on the day of the former heathen feast of the Sun.
While 5G NR is being specified, and 5G networks deployed, the 5G ecosystem
will be enriched by additional ingredients. One of the most relevant is the adoption of
Artificial Intelligence (AI) technology for the extraction of useful information from
the huge amounts of data that IoT networks will generate. AI will be a key ingredient
for the concept of "digital twin", intimately connected to the IoT.
2 Not Only 3GPP RATs
In the context of IoT, nowadays manufacturers and stakeholders are looking favorably
in the direction of Low Power Wide Area (LPWA) systems. These technologies are
devised, as their name suggests, to achieve both longer coverage ranges and low
energy consumption. These conditions are highly effective for sensors and actuators
networks in IoT, since there is the need of many devices per unit area and no need of a
power grid in the vicinity. Maintenance and deployment costs are then reduced. LTE-
M, NB-IoT and LoRa can be classified as LPWA technologies, where the first two
IoT Support 191
are bounded to the cellular network and the last one can be deployed as standalone.
Note that also the EC-GSM-IoT is a cellular technology for the IoT. However, it
leverages on GSM/GPRS frequency bands and it is not expected to keep a relevant
portion of the market in the way to 5G. For these reasons it will not be discussed in
technical details here.
At the same time, 5G New Radio (NR) and C-V2X are 3GPP RATs that can have
a stray impact on specific IoT domains.
2.1 5G New Radio
As foreseen in the previous years, 5G will include deployment scenarios of var-

ious type, especially in Enhanced Mobile Broadband (eMBB), Massive Machine
Type Communication (mMTC), Ultra-Reliable and Low Latency Communications
(URLLC), and Enhanced Vehicle to Everything (eV2X). The IoT context falls within
mMTC and partly in URLLC scenarios.
However, 5G NR has scalable configurations to allow more efficient handling
of mMTC and URLLC applications. For example, it supports for up-link non-
orthogonal multiple access scheme, and, for the scheduling of resources, an up-link
transmission scheme without grant is possible. Moreover, the criticality of URLLC
links requires additional flexibility in the definition of slots and mini-slots to reduce
the end-to-end latency. Therefore, 5G NR is designed to also interface with the
always increasing traffic demand coming from IoT devices.
2.2 NB-IoT and LTE-M
An increasing interest in cellular technologies for IoT (or massive MTC) paves the
way to the introduction of Narrowband IoT (NB-IoT) and LTE for Machines (LTE-M
or LTE CatM1).
NB-IoT is designed for an efficient communication and longer battery life for
massively distributed nodes. Its three key elements are the low costs, a large number
of connections per cell and an optimal choice for coverage, with very good penetration
in indoor environments and underground [3].
Technically, NB-IoT took its first standardization in 3GPP Release 13 and therefore
it takes the LTE numerology for synchronization, radio access, resources definition
and assignment. Of course, it employs modifications with respect to LTE UEs to allow
longer coverage range and low energy consumption. This comes at the expense of
lower data rates.
While the other cellular systems for MTC are based on existing radio access
technologies, NB-IoT could be considered a "new" radio access technology. In fact,
it can either operate in a stand-alone mode or within the guard bands of LTE carriers
or within an LTE carrier. Being narrow-band, it supports a nominal system bandwidth
of 180 kHz with a channel spacing that can be decreased to 3.75 kHz for the uplink.
Thus, it guarantees a higher spectrum flexibility and system capacity with respect to
other technologies. Furthermore, it allows energy efficient operation thanks to the
possible configuration of three classes of devices based on power levels, by keeping
an ultra-low device complexity. NB-IoT becomes then very competitive in the IoT
market.
Along with NB-IoT, LTE-M takes the advantages of being a LPWA technology
with an LTE base. LTE-M offers a lower coverage range with respect to NB-IoT, but
it also proves to reach higher peak data rates, four times the peak rates of NB-IoT.
However, LTE-M is mainly dedicated to employ an efficient power consumption
and gain from reduced device complexity. Still, its coverage range is extended with
respect to the other UE categories of LTE. LTE-M was originally designed as an LTE-
based technology competitive with EGPRS in the IoT market. LTE-M is a LPWA
technology defined in the Release 13 of 3GPP specifications, specifically as LTE
CatM1. LTE-M can co-exist with 2G, 3G and 4G mobile radio networks and benefit
from the privacy and security options which are available and previously designed
for them. Compared to others like NB-IoT, LTE-M is optimized for higher bandwidth
and mobile (in the sense of moving nodes) connections. As a consequence, it reaches
up to 1 Mbps of data rate and well supports nodes mobility. One clue advantage of
LTE-M included in the standard, is the possible choice for operation in either full-
duplex FDD, half duplex FDD or time division duplex (TDD). Its modulation is
still OFDMA, following LTE numerology. One other peculiarity of LTE-M is the
support of voice over LTE. Furthermore, for what concerns power saving, a longer
battery life should last about 10 years for the IoT devices, with reduced modem costs
of 20-25% of the current EGPRS modems [10].
For more technical details and comparison between cellular technologies, see
Table 1.
Table 1 Comparison among LTE-M and NB-IoT

Technology LTE-M NB-IoT
Deployment In-band LTE In-band & Guard band LTE, stan-
dalone
Bandwidth 1.08 MHz 180 kHz
UL access SC-FDMA, 15 kHz subcarrier SC-FDMA, 15 or 3.75 kHz sub-
spacing, Turbo code carrier spacing, Turbo code
DL access OFDMA, 15 kHz subcarrier spac- OFDMA, 15 kHz subcarrier spac-
ing, Turbo code ing, TBCC
Duplexing FD & HD (type B), FDD & TDD HD (type B), FDD
Coverage 155.7 dB 164 dB for standalone
Peak rate 1 Mbps in UL & DL 250 kbps in UL for multi-tone and
250 kbps in DL
Power class 23 or 20 dBm 23 dBm
Power saving PSM, ext. I-DRX, C-DRX PSM, ext. I-DRX, C-DRX
IoT Support 193
2.3 C-V2X
Vehicular networks are also part of the IoT scenario. Vehicles themselves can be
nodes of a vehicle sensor network. Transportation might become a mean not only to
transmit safety data to other vehicles, but also a mean to increase wireless coverage
and move data to different areas. Furthermore, it also fosters use cases for other
traffic participants, like cyclists and pedestrians [7].
The novel Cellular Vehicle-to-Everything (C-V2X) standard is based on 4G, and
differently from earlier technologies that are based on IEEE 802.11p, can support
a wider set of capabilities [9]. The first example are direct communications for
V2V, V2I V2Pedestrians. C-V2X operates in the 5.9 GHz frequency band and it is
capable of working independently of cellular networks. It is agreed worldwide to
have dedicated frequencies not subject to interference and exploit direct low latency
links within short distances. Bandwidths can be of 10 or 70 MHz, depending on
whether basic or advanced safety services are requested.
Of relevant note is the fact that by operating in the device-to-device mode (as
for V2V, V2I, V2P), it is not mandatory for C-V2X to rely on any network infras-
tructure [6]. C-V2X will allow vehicles to support an interesting set of features, like
cooperate, coordinate and share information collected by sensors (comparable to a
wireless sensor network for the IoT) in advanced driver assistance systems, as well
as connected automated driving. Moreover, the use cases identified by the 5G Infras-
tructure Association can be subdivided in two architecture types [8]: one with lower
tolerance to errors and up to 1 Mb/s of data rate, and the other with a more relaxed
tolerance but requesting tens of Mb/s. Both of them require end-to-end latency in
terms of few or few tens of ms.
C-V2X must be fully compatible with 5G by design and expanded in Release 15
and Release 16. Thanks to the increased quantity of data available in 5G, vehicles
will be able to collect more information. A wide range of business models is then
made available, ranging from telematics, infotainment, to real-time mapping, and
data analytics.
2.4 LoRa
IoT standards are not bound to 3GPP. Because of the massive number of possible IoT
applications, many technologies can operate alongside 5G. For example, the LoRa
technology gained its success for its simplicity and efficiency in the WSN context.
It is worth of mention because it is highly likely that system based on LoRa will
probably have to interface with 5G slices.
LoRaWAN is one of the first technologies defined for LPWAN applications. It
specifies a MAC and PHY layer, and we refer to the latter simply as LoRa. Its
standardization was initiated by Semtech, an American company, and later by the
LoRa Alliance, an organization of companies.
LoRa exploits ISM bands, and it is very robust with respect to the interferers
present in the same shared band. LoRa is based on a proprietary modulation based on
chirp spread spectrum, with pulses whose frequency increases or decreases linearly
over time; information is inserted in these pulses by introducing a discontinuity at
different time offsets from the start of a symbol. Bandwidth can be 125, 250 or
500 kHz. Interference is mitigated with the use of forward error correcting codes
in combination with frequency hopping spread spectrum. An important parameter
for the PHY layer is the Spreading Factor (SF), that is the ratio between the signal
bandwidth and the symbol rate. An increase of the SF of one unit, in a range from 6
to 12, corresponds to a doubling of the time on air (duration of a packet transmission)
and a decreasing of the receiver sensitivity of roughly three dB.
Then, LoRaWAN introduces three classes of devices with differences in the MAC
layer to allow complexity and energy consumption dependent on the application.
3 Application Domains
5G networks will support the IoT through their RATs in a variety of application
domains. This section shortly introduces those who are promising to attract more
interest of major industry stakeholders.
3.1 Industry 4.0
A term invented in Germany in 2011 to represent the 4-th industrial revolution,

Industry 4.0 includes many different application domains; most often, it is used
to indicate Smart Manufacturing approaches where wireless communications and
cyber-physical technologies are applied to industry plant automation.
3.2 Smart Manufacturing
The application of 5G radio technologies to industry plants might introduce a num-

ber of benefits to automation systems, as long as the stringent requirements set in
terms of reliability and latency will be met. In particular, making wireless the links
between sensors and actuators on robotic machines, will simplify their maintenance
and design, and will permit to add monitoring devices on components currently
unreachable because of the impossibility to deploy wires. While the due levels of
link reliability might be achieved through the application of proper transmission
techniques, the requirements on maximum latency might still be a limit; some in-
dustrial applications require control loops with maximum delays in the order of
tens of microseconds, a level unreachable even by 5G NR. On the other hand, wire
IoT Support 195
replacement is a significant advantage, and for those applications where latency is

not an essential requirement, 5G will be an enabler of increased process efficiency.
Which RAT will best fit to the user needs is difficult to predict; 5G NR promises to
deliver low-latency high reliability services, but at the cost of implementation of a
SIM card in UEs. Will the industry managers accept this step?
3.3 Protection of Civil Infrastructures
The adoption of IoT technologies, connecting machines to the Internet, allows in-
dustry to implement the so-called "digital twin" paradigm: (physical) factory plants
are accompanied by a (virtual) twin, which describes the mechanics of machines and
provides an updated status of the running processes. The more sensors are deployed
on the plants, measuring physical parameters of all types and transmitting them in
real time to the digital twin, the stronger is the ability of algorithms to detect poten-
tial future failures and to predict optimally the time for intervention of maintenance
operators.
The same concept can be applied for the sake of protection of civil infrastructures,
like e.g. bridges, roads, dams. The digital twins of such physical entities might be
used to predict and avoid potential collapses, or damages. This would require models
able to represent the behavior of the infrastructure. Unfortunately in this case these
models are often incomplete or unknown. Under such circumstances, approaches
based on AI might be used to overcome the lack of models.
The adoption of the concept of digital twin in civil engineering would requires
huge amounts of data generated and transmitted by sensors deployed on and in-
side infrastructures. On the other hand, requirements on latency and reliability of
transmissions might be not stringent, as the AI algorithms would run under a sort of
mid-term perspective. So, the most relevant requirement would be on the overall data
rate generated by any individual infrastructure (embedding hundreds of IoT nodes);
for such application, 5G NR is a must-have technology.
3.4 Smart Agriculture
Tens or hundreds of sensors per hectare will be deployed in agriculture to monitor

the health of vineyards, olive trees or other types of cultivations. Most applications
under this category do not pose stringent requirements on latency. The traffic density
is very low despite the large numbers. The energy efficiency of the RAT used must
be extremely high, to ensure long lifetime for devices embedded in terrain or on
trees. LoRaWAN seems to be the most efficient technology.
3.5 Smart Cities
While LoRaWAN is currently deployed in many cities for the provision of remote
metering services (and other), there is a general thread: as long as the traffic generated
by IoT devices in cities will increase significantly, LoRaWAN network might saturate
owing to the lack of reserved frequency bands and the long ranges covered by
LoRaWAN gateways. This will make room to NB-IOT and/or LTE-M success.
3.6 Automotive Applications
For some years automotive manufacturers have been wondering about whether to
rely on ad-hoc networking approaches (like WAVE, based on 802.11p), or wait for 5G
deployment. Now, C-V2X promises to enable most of services needed for connected
cars. It is most likely that C-V2X will soon become the standard communication sys-
tem for automotive applications, though advanced services like augmented sensing
might require 5G NR interface.
3.7 Unmanned Aerial Vehicles (UAVs)
Soon the volume of space above densely populated areas will be the new frontier
for radio technology. Using mobile radio networks to coordinate the flight of UAVs
patrolling cities from the sky, has become a clear interest of many stakeholders;
the limit is still in regulations, which will evolve as long as safety threads will be
faded away through the application of proper technologies on UAVs. However, it
is envisioned that the evolution of 5G will include the use of mobile base stations,
carried by UAVs. In this case, their control will require highly reliable links (easy
to achieve in Line-Of-Sight conditions) and very low latencies, compatible with 5G
NR evolution.
4 Expected Trends
Based on the analysis reported above, the 5G ecosystem will optimally serve the
various IoT application domains with different RATs. Mobile Network Operators
will need to be able to offer services based on a multi-RAT approach comprising
non-3GPP solutions. Among the latter ones, which use ISM bands, LoRaWAN is
becoming more and more successful; it works mostly on the 868 MHz band, though
recent releases of LoRa chipsets operating at 2.4 GHz will make the adoption of this
frequency band feasible. In any case, its use for long-range applications in densely
populated areas might encounter problems in terms of saturation of the frequency
IoT Support 197
bands. One option to solve this issue might rely on the identification of separate
frequency bands specifically for smart city scenarios (as done for other application
domains, like e.g. for health).
In any case, what the 5G ecosystem might bring as strong support tool to the IoT
world, lies mostly in the cloud computing component, the adoption of AI approaches,
the development of digital twin technologies. MNOs will work in that direction. What
RAT is used by things, should be transparent to such elements of the ecosystem.
Nevertheless, the availability of different RATs, each one being optimal for a different
IoT application domain, is an essential aspect of the 5G ecosystem.
References
1. "Towards a definition of the Internet of Things (IoT)", IEEE Internet Initiative, May 2015.
2. "Service requirements for the 5G System; Stage 1", Technical Specification 3GPP TS 22.261
V16.4.0, June 2018.
3. Olof Liberg, former 3GPP GERAN and GERAN WG1 Chairman (Oct. 24, 2017) "The Cellular
Internet of Things" http://www.3gpp.org/news-events/3gpp-news/1906-c_iot
4. 3GPP TR 36.888, "Study on provision of low-cost Machine-Type Communications (MTC)
User Equipments (UEs) based on LTE", 3GPP
5. 3GPP TR 38.912, "Study on new radio (NR) access technology", 3GPP Release 14
6. 5G Automotive Association, "The Case for Cellular V2X for Safety and Cooperative Driv-
ing", White Paper, November 16, 2016 http://5gaa.org/news/white-paper-placeholder-news-
for-testing/
7. NGMN Alliance, "V2X White Paper", White Paper, June 17, 2016
https://www.ngmn.org/fileadmin/ngmn/content/downloads/Technical/2018/V2X_white_paper_v1_0.pdf
8. 5G Infrastructure Association, "5G Automotive Vision", White Paper, October 20,
2015 https://5g-ppp.eu/wp-content/uploads/2014/02/5G-PPP-White-Paper-on-Automotive-
Vertical-Sectors.pdf
9. GSMA, "Cellular-Vehicle-to -Everything (C-V2X) Enabling Intelligent Transport"
https://www.gsma.com/iot/wp-content/uploads/2017/12/C-2VX-Enabling-Intelligent-
Transport_2.pdf
10. GSMA, "Long Term Evolution for Machines: LTE-M" https://www.gsma.com/iot/long-term-
evolution-machine-type-communication-lte-mtc-cat-m1/
The Role of Cloud and MEC in 5G
Fabrizio Granelli
Abstract 5G will incorporate several state-of-the-art architectural and protocol ap-

proaches on top of his networking infrastructure to tackle the emerging needs for
improved flexibility and performance.
Such requirements lead to the design of two options for the 5G architecture: one,
which represents an evolution of 4G LTE standard IP architecture, and the other
where the core network functions interact with each other using a Service Based
Architecture (SBA). This novel architecture allows the integration of the recent de-
velopments in the field of Cloud technology and Mobile Edge Computing. This
chapter will introduce how the cloud approach and edge computing paradigm can
be integrated in the upcoming 5G standard for next generation cellular networks.
1 Introduction
5G is expected to support unprecedented requirements. Indeed, besides a predictable

increase in data transfer performance and spectral efficiency, 3GPP 5G requirements
include very low latency (in the order of msec), high reliability, capability to offer
access to distributed computation and storage facilities in addition to connectivity
and bandwidth.
Knowing the complexity of current 4G/LTE architectures, it seems extremely hard
to define a single architecture able to satisfy ALL the expected performance require-
ments. Nevertheless, several target scenarios and services include strict constraints
in terms of low latency and extremely reliability. Those services can be classified as
URLLC (Ultra-Reliable Low Latency Communication) services. Vehicular commu-
nications and remote control of robots or machinery belong to such class of services,
and they are well-known 5G application scenarios.
Fabrizio Granelli
Consorzio Nazionale Interuniversitario per le Telecomunicazioni - Sede di Trento, Via Sommarive
9, Trento, ITALY, e-mail: fabrizio.granelli@unitn.it
199
200 Fabrizio Granelli
As a consequence of the above issues, it is necessary for the 5G architecture to

incorporate additional functionalities, by exploiting the recent solutions existing on
the Internet: the Cloud and virtualization.
Cloud Computing represents the current paradigm for the delivery of services to
a massive number of users, and it is based on the concentration of huge computation
and storage resources in strategic locations (the datacenters) that can run services in
an efficient and scalable manner by using advanced management and virtualization
approaches.
Virtualization represents a relatively old paradigm for enabling access to shared
resources, that in the modern days gained enormous attention due to its successful
application in support of cloud computing, but more recently also in the area of
networking. Indeed, by virtualizing key network functionalities, it is possible to
detach software functions from dedicated hardware, with the advantage to be able to
re-locate or modify their resources most efficiently and in real-time. This emerging
paradigm is defined Network Function Virtualization (NFV).
The above concepts can be introduced in the design of the next generation of
mobile networks, leading to the definition of two emerging paradigms in the design
of 5G: Cloud Radio Access Network (Cloud RAN) and Mobile Edge Computing or
Mobile Edge Cloud (MEC). The purpose of this chapter is to provide some basic
information of such paradigms as well as their relationship with the 5G standard.
2 The Cloud Radio Access Network paradigm
A relevant issue in the design of next generation mobile networks is related to the
fact that, in order to enable the wireless communication technologies to provide
high performance to the mobile users, cells are increasingly becoming smaller and
smaller. This generates problems in terms of costs for buying the equipment as
well as powering them, since currently (in 4G) cellular Base Stations are extremely
expensive.
In this framework, the Cloud RAN (C-RAN), sometimes also referred to as
Centralized-RAN, represent a possible architecture for future cellular networks to
tackle the issues of costs and energy consumption [1].
Indeed, Base Stations are designed to handle the maximum traffic, not average
traffic, resulting in a waste of processing resources and power at idle times or in
situations of low amount of traffic. The problem is related to the fact that the majority
of power does not scale with the number of users or the amount of traffic, being related
to power the RF interfaces and cooling the processing elements. Therefore, a more
flexible solution is required.
The typical RAN architecture in 4G/LTE is depicted in Figure 1, where the
Base Stations contain both RF and signal processing functionalities in the so called
"evolved Node B" (eNodeB) architecture. Cloud RAN introduces a detachment
between the RF interfaces (called Remote Radio Heads - RRHs) and the signal
and data processing functionalities (called Base Band Units - BBUs), leading to an
The Role of Cloud and MEC in 5G 201
Fig. 1 Typical RAN Architecture: the eNodeB contains both RF and processing capabilities.
architecture like the one presented in Figure 2. To some extent, C-RAN may be viewed
as an architectural evolution of a distributed base station system, that takes advantage
of many technological advances in wireless, optical and IT communications systems.
Moreover, Cloud RAN is a paradigm whose scope is limited to the 5G Radio Access
Network (RAN), and especially is impacts mostly the eNodeB architecture.
In Cloud RAN, the detachment of BBUs from the RRHs is achieved by means
of the latest CPRI standard, that enables to interconnect the two components us-
ing optical fibers or wireless through the fronthaul links. Optical trnasmission is
achiwved through Coarse or Dense Wavelength Division Multiplexing (CWDM/
DWDM) technology, while mmWave communications are used to allow transmis-
sion of baseband signal over long distance.
BBUs are then grouped together by applying recent Data Centre Network technol-
ogy to allow a low cost, high reliability, low latency and high bandwidth interconnect
network within the BBU pool. There, open platforms and real-time virtualisation
technology rooted in cloud computing are employed to achieve dynamic shared
resource allocation and support of multi-vendor, multi-technology environments.
Figure 3 provides a visual representation of the functional split between BBU (on
the left) and RRH (on the right, in green) functions.
The Cloud RAN enables large scale centralized deployments: it allows hundreds
of thousands of remote RRHs connect to a single centralized BBU pool. In this
scenario, latency limits indicated in the standards play a major role in the design and
dimensioning of the system and in the degrees of freedom in positioning the BBU
pools. For this reason, the maximum distance can be 20 km using an optical fiber link
for 4G (LTE/LTE-A) system, but longer distance (40 km 80 km) is possible for 3G
(WCDMA/TD-SCDMA) and 2G (GSM/CDMA) systems. As an example of actual
deployments, some Asia operators claim to have deployments of C-RAN systems
with 1200 RRHs centralized to one central office.
Fig. 2 The Cloud RAN Architecture concept: BBU is detached from RRH and move into a BBU
pool.
Fig. 3 Functional splitting between BBU and RRH. BBU and RRH communicate through the
CPRI interface.
Another advantage of Cloud RAN is the native support to collaborative radio

technologies. Indeed, using the C-RAN architecture, any BBU can talk with another
BBU within the BBU pool with very high bandwidth (10Gbit/s and above) and
extremely low latency (10us level). This is enabled by the interconnection of BBUs
within the pool.
It should be noted that Cloud RAN is different from BBU Hoteling, or Base
Station Hoteling. While in Cloud RAN BBUs can also be interconnected among
them and flexibly re-located (as they practically Virtual Machines or containers),
in hoteling the BBUs of different base stations are simply stacked together and no
direct link among them is available.
Recently, a mathematical framework was developed at the CNIT Research Unit
in Trento to study the potential benefits of Cloud RAN solutions in the framework
of 5G. By using the model developed in [2], the team demonstrated the flexibility
of the Cloud RAN architecture to adapt to the daily variations of the traffic pattern
of a typical mobile operator. Moreover, they studied the possibility to consolidate
the BBU VMs in order to provide further resource savings. Figure 4 provides an
example of the achieved results, where the energy gain is analyzed for different BS
configurations and at different hours of the day.
Please check [1] for further details on Cloud RAN and [2] for more details on the
mathematical model.
Fig. 4 Potential energy gains deriving from the deployment of the Cloud RAN paradigm: in
presence of low traffic, fewer BBUs are required.
3 Mobile Edge Computing
Multi-access Edge Computing (MEC), or Mobile Edge Computing, is a network

architecture concept that enables cloud computing capabilities and an IT service
environment at the edge of the cellular network and, more in general at the edge
of any network. The basic idea behind MEC is that by running applications and
performing related processing tasks closer to the cellular customer, network conges-
tion and latency are reduced and as a consequence applications can provide better
performance. This is especially true in case of applications requiring low latency
from the network, such as automated vehicle driving, remote operation of devices,
etc.
MEC technology is designed to be implemented at the cellular base stations or
other edge nodes, and enables flexible and rapid deployment of new applications and
services for customers. Nevertheless, in order to deploy MEC solutions in an effective
manner, cellular operators are required to open their radio access network (RAN) to
authorized third-parties, such as application developers and content providers. This
represents an additional requirement in the design of the 5G architecture.
The 5G system architecture specified by 3GPP and described in [3] is designed to
address a wide set of use cases, which can be typically clustered into three groups:
Enhanced Mobile BroadBand (eMBB), massive Machine Type Communications

(MTC), and Ultra-Reliable Low Latency Communications (URLLC).
As discussed on the introductory section, supporting all use cases with a common
architecture has required significant changes in design philosophies both for the
RAN and the core network. In the 5G system specification there are two options
available for the architecture: one with the traditional reference point and interface
approach, which represents an evolution of 4G LTE standard IP architecture, and
the other where the core network functions interact with each other using a Service
Based Architecture (SBA). Indeed, the SBA represents a big step forward in the
virtualization and softwarization of the architecture.
Details on the SBA option of the 5G system architecture are provided in a white
paper by ETSI [4]. In brief, the SBA framework is built around functions, that con-
sume services and/or produce services. Any network function can offer one or more
services. The SBA framework provides the necessary functionality to authenticate
the consumer and to authorize its service requests, as well as flexible procedures to
efficiently expose and consume services. For simple service or information requests,
a request-response model can be used. For any long-lived processes, the framework
also supports a subscribe-notify model.
ETSI Industry Specification Group on Multi-access Edge Computing (ETSI ISG
MEC) defined an API framework aligned with the above principles. The functionality
needed for efficient use of the services includes registration, service discovery,
availability notifications, de-registration and authentication and authorization. All
this functionalities are the same in both the SBA and the MEC API frameworks.
Fig. 5 MEC deployment in the 5G architecture proposed by ETSI.
The network functions and the services they produce are registered in a Network
Resource Function (NRF), while in MEC the services produced by the MEC ap-
plications are registered in the service registry of the MEC platform. 5G Network
Exposure Function (NEF) acts as a centralized point for service exposure and also
has a key role in authorizing all access requests originating from outside of the
system.
One of the key concepts in 5G is Network Slicing, that allows the allocation of
the required resources from the available network functions to different services or
to tenants that are using the services. The Network Slice Selection Function (NSSF)
is the function that assists in the selection of suitable network slice instances for
users, and in the allocation of the necessary Access Management Functions (AMF).
A MEC application, i.e. an application hosted in the distributed cloud of a MEC
system, can belong to one or more network slices that have been configured in the
5G core network.
The Unified Data Management (UDM) function is responsible for generating the
3GPP AKA authentication credentials, handling user identification related informa-
tion, managing access authorization (e.g. roaming restrictions), registering the user
serving NFs (serving AMF, Session Management Function (SMF)), supporting ser-
vice continuity by keeping record of SMF/Data Network Name (DNN) assignments
and performing subscription management procedures.
The User Plane Function (UPF) has a key role in an integrated MEC deployment
in a 5G network. UPFs can be seen as a distributed and configurable data plane
from the MEC system perspective. The control of that data plane, i.e. the traffic
rules configuration, now follows the NEF-PCF-SMF route. Consequently, in some
specific deployments the local UPF may even be part of the MEC implementation.
The resulting integrated architecture described in the white paper is presented in
Figure 5, where the 3GPP 5G SBA system is shown on the left, while the ETSI MEC
architecture is on the right.
The MEC system is presented on the right-hand side of Figure 5. The core module
of the MEC system is the MEC orchestrator, a MEC system level functional entity
that, acting as an AF, can interact with the Network Exposure Function (NEF). In
some scenarios, the MEC orchestrator can also interact directly with the target 5G
NFs. On the MEC host level it is the MEC platform that can interact with these 5G
NFs, again in the role of an AF. The MEC host, i.e. the host level functional entities,
are most often deployed in a data network in the 5G system.
Figures 6-9 depicts the available physical deployment alternatives for the deploy-
ment of MEC within 5G:
1. MEC and the local UPF co-located with the Base Station (Fig. 6).
2. MEC co-located with a transmission node, possibly with a local UPF (Fig. 7).
3. MEC and the local UPF co-located with a network aggregation point (Fig. 8).
4. MEC co-located with the Core Network functions (i.e. in the same data centre,
Fig. 9).
Basically, the architecture enables to deploy MEC in different locations between
the Base Station and a remote Data Center. Nevertheless, all deployments have
in common the UPF that is used to steer the traffic towards the targeted MEC
applications and towards the network. It should be noted that the MEC management
Fig. 6 Possible physical deployment of MEC in 5G networks - co-located with the Base Station.
Fig. 7 Possible physical deployment of MEC in 5G networks - co-located with a transmission node.
system, which orchestrates the operation of MEC hosts and applications, may decide
dynamically where to deploy the MEC applications.
4 Conclusions
Cloud RAN and MEC represent a big step forward in increasing the flexibility of
the 5G network infrastructure, with the purpose to improve scalability and service
support.
This chapter provided a brief introduction to the above concepts, providing some
discussion on the potential benefits of such emerging paradigms and their integration
in the upcoming 5G Standard.
Fig. 8 Possible physical deployment of MEC in 5G networks - co-located with a network aggrega-
tion point.
Fig. 9 Possible physical deployment of MEC in 5G networks - within the same data center.
References
1. A. Checko and H. L. Christiansen and Y. Yan and L. Scolari and G. Kardaras and M. S.
Berger and L. Dittmann, "Cloud RAN for Mobile Networks—A Technology Overview",
IEEE Communications Surveys Tutorials, Vol. 17, No. 1, pp. 405–426 (2015), doi:
10.1109/COMST.2014.2355255
2. R. Bassoli, M. Di Renzo and F. Granelli, "Analytical energy-efficient planning of 5G cloud
radio access network," 2017 IEEE International Conference on Communications (ICC), Paris,
pp. 1–4 (2017), doi: 10.1109/ICC.2017.7996871
3. 3GPP TS 23.501 V15.1.0, "3rd Generation Partnership Project; Technical Specification Group
Services and System Aspects; System Architecture for the 5G System; Stage 2 (Release 15)"
(2018-03).
4. ETSI White Paper No. 28, "MEC in 5G networks," First edition, June 2018, ISBN No.
979-10-92620-22-1
5G for V2X Communications
Antonella Molinaro and Claudia Campolo
Abstract Connected and fully automated vehicles are expected to revolutionize

transportation systems of the future on a global scale, significantly improving road
safety and traffic efficiency, and fostering investments in the automotive market.
This new landscape mandates a robust, flexible and business agile communication,
networking and computing technology foundation, in which fifth generation (5G)
systems and enablers will take the lion’s share. By leveraging new-designed efficient
air interfaces, a wide range of allocated frequencies, advanced transceivers, multiple
radio access technologies, as well as cutting-edge network softwarization princi-
ples, 5G intends to guarantee ultra-low latency, ultra-high reliability, and high-data
rate vehicle-to-everything (V2X) connectivity. This chapter will describe the status
quo of V2X communications, by analyzing the main application requirements and
the restless activities engaging the R&D community towards 5G-enabled V2X. Re-
search challenges and opportunities, along with perspectives about system design
and enablers for 5G-V2X communications will also be part of this chapter.
1 Introduction
The automotive vertical market is undergoing key technological transformations,

with the focus of innovation shifting towards connected and fully automated (i.e.,
autonomous or driverless) vehicles. The potential of such transformation will be
Antonella Molinaro
University Mediterranea of Reggio Calabria, Loc. Feo di Vito, Reggio Calabria, Italy, e-mail:
antonella.molinaro@unirc.it
Laboratory of Signals and Systems, CentraleSupélec-CNRS-University Paris Sud, University Paris
Saclay Gif-sur-Yvette, France, e-mail: antonella.molinaro@l2s.centralesupelec.fr
Claudia Campolo
University Mediterranea of Reggio Calabria, Loc. Feo di Vito, Reggio Calabria, Italy, e-mail:
claudia.campolo@unirc.it
209
210 Antonella Molinaro and Claudia Campolo
fully disclosed when independent autonomous vehicles will become connected and
cooperate with one another, with roadside infrastructure elements, with pedestrians
and other vulnerable road users (VRUs), and with the cloud through vehicle-to-
everything (V2X) communications. V2X can create a vehicle’s collective perception
of the surrounding environment and help it making more informed decisions, based
on exchanged local views and planned manoeuvres from nearby vehicles, instead of
relying on local awareness built upon on-board sensors only (e.g., radar, LIDAR,
cameras) like an autonomous vehicle would do.
Connected and fully automated vehicles will combine to bring about safer trans-
portation aiming at zero fatalities on the road, improved traffic flow with the support
of the roadside infrastructure, and consequent low environmental impact. However,
the complexity of this landscape raises unprecedented challenges. V2X applications
such as cooperative sensing and maneuvering, high-density platooning, tele-operated
driving show hard-to-meet computing and communication demands, well beyond
what the current radio access technologies are able to provide today. Ultra-low la-
tency (below 10 ms), ultra-high reliability (near 100%) and high data rate (in the
order of Gbps) communications are demanded by most V2X safety applications.
In addition, the inherent dynamics in vehicular environments related to the rapidly
changing network topology, the fast-varying wireless channel and possible intermit-
tent connectivity, further increase the system design complexity and overall need
an end-to-end comprehensive approach. An optimal end-to-end chain of applica-
tions and (edge/cloud) services, radio access and core network functionalities is
required to tackle V2X demands and challenges and to maximise the benefits of
future investments in the automotive market.
There is a general consensus among the stakeholders, as confirmed by a plenty
of initiatives, about the role of 5G as the game changer to realize such a challenging
vision. The Third Generation Partnership Project (3GPP), after releasing specifi-
cations for cellular V2X (C-V2X) in Release 14 and 15, is currently discussing
further enhancements to the 5G architecture in Release 16 [2], in order to meet the
most demanding V2X performance requirements. The 5G Automotive Association
(5GAA) [5], formed in September 2016 by the major automobile manufacturers
and telco players, promotes interoperable end-to-end 5G-based V2X connectivity. A
similar intent is shared by the 5G Infrastructure Public Private Partnership (5G-PPP)
[26], with projects underway that target such ambitious goals, such as 5GCAR [7]
and 5GCARMEN [8]. Also governments around the world are supporting or even
advocating connected and automated vehicles [19], [20].
In such a context, this chapter aims to provide an overview of the ongoing efforts
in the race towards enabling V2X communications in 5G systems. Solutions for
V2X use cases support will be analyzed from an evolutionary point of view, by
first scanning early and current enabling technologies, and then investigating future
opportunities offered by 5G solutions to the automotive vertical market.
5G for V2X Communications 211
Fig. 1 V2X communication modes, interfaces and entities.
2 Status quo of V2X communications
2.1 Pre-5G radio access technologies
IEEE 802.11. IEEE 802.11, in particular its amendment .11p (ITS-G5 in Europe),
has been investigated since nearly two decades as the enabling radio access technol-
ogy for Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communica-
tions. Its main attractive feature was the capability of supporting distributed localized
interactions among vehicles even in the absence of a roadside infrastructure. Several
worldwide field trials have demonstrated the .11p feasibility of supporting coopera-
tive awareness applications (e.g., emergency brake light, stationary vehicle warning),
and truck platooning [13]. Although .11p fits the requirements of such applications
in low congested scenarios, it suffers from dramatic throughput degradation and
poor performance at high density conditions, so it cannot match the very low latency
and high-bandwidth requirements of future V2X applications. Such limitations are
mainly due to a very basic physical layer and to the lack of a protection mechanism
from interference and collisions – which is especially critical for broadcast com-
munications under congestion – as ruled by the distributed Medium Access Control
(MAC) operating with the Carrier Sense Multiple Access with Collision Avoidance
(CSMA/CA) protocol [14]. The mandate and rulemaking for 802.11-equipped cars
started by US government some years ago has been frozen recently in front of the
appearance of the cellular technology as another candidate for V2X support.
3GPP Cellular V2X (C-V2X). The role of 3GPP and cellular networks for V2X
has been steadily and rapidly growing, with C-V2X as part of the 3GPP Long Term
Evolution (LTE) program, the first stage of specifications for Release 14 completed
in June 2017, and a clear roadmap towards further refinements in Release 15 and
enhanced capabilities expected for 5G in Release 16. The C-V2X rollout can be
facilitated by the ubiquity of the cellular infrastructure, its centralized organization
and mature industrial foundation. Furthermore, 3GPP inherits the results of decades
of previous standardization works in other organizations that defined vehicular ap-
plications and messages [14].
In 3GPP documents, the term V2X collectively refers to communications among
different entities, as illustrated in Fig. 1: (i) V2V for direct communication between
vehicles in close proximity; (ii) V2I for communication between vehicles and a
roadside unit (RSU) in radio range, which can be implemented either in an eNodeB or
in a standalone device (e.g., a traffic light); (iii) Vehicle-to-Pedestrian (V2P) between
vehicles and vulnerable road users (e.g., pedestrians, bikers); and (iv) Vehicle-to-
Network (V2N) for communications with remote servers and cloud-based services
reachable through the cellular infrastructure.
V2V communications, considered as the highest priority for 3GPP, required mod-
ifications in the radio access network [3]. Efforts resulted in the specifications of two
communications modes, namely Mode 3 (scheduled) and Mode 4 (autonomous),
supporting direct communications over the sidelink (PC5) interface. Communica-
tions on the PC5 interface use the 5.9 GHz band, independent or even in the absence
of a cellular network, in order to ensure ultra-high availability under all geographies,
regardless of the specific Mobile Network Operator (MNO). In Mode 3, operating
only in-coverage of an eNodeB conditions, the allocation of radio resources is super-
vised by the network; whereas in Mode 4 pre-configured resources can be accessed
by vehicles in an autonomous manner without the network control, both in- and
out-of coverage of an eNodeB (e.g., in urban canyons, tunnels) [21]. Early results
have demonstrated the superior performance of C-V2X Mode 4 w.r.t. IEEE 802.11p
under many circumstances [21], [23].
V2N communications occur over the cellular LTE-Uu interface operating in the
traditional licensed spectrum. Less disruptive although not negligible modifications
have been applied to this interface in order to support both unicast and multicast
communications.
Architectural enhancements have been also specified to support Vehicular User
Equipments (VUEs) and to manage V2X communications [4]. Notably, this entailed
adding the V2X Control Function module, which provides configuration parameters
for VUEs located in and out-of-coverage of an eNodeB, and the V2X Application
Server (V2X AS), responsible for the control and distribution of traffic, road and
service information.
2.2 V2X applications and requirements
V2X applications cover a wide range of use cases, which can be clusterized based on
their purpose and requirements [19]. The 5GAA has grouped V2X use cases in four
categories: Safety aimed at reducing the frequency and severity of vehicle collisions;
Convenience managing the vehicle health and offering services like diagnostics and
software updates; VRU targeting safe interactions between vehicles and non-vehicle
road users; Advanced driving assistance sharing similar objectives with the safety
Table 1 Requirements of V2X autonomous driving use cases [1].

Application Main commu- Payload Latency Reliability Data rate
nication mode (bytes) (ms) (percentage) (Mbps)
Vehicles platooning V2V, V2I 50-6500 10-20 90-99.999 0.012-65
Advanced driving V2V, V2I 300-12000 3-100 90-99.999 0.096-53
Extended sensors V2V, V2I, V2P 1600 3-100 90-99.999 10-1000
Remote driving V2N - 5 99.999 25 (Uplink);
1 (Downlink)
use cases, but treated separately for their close relationship with (semi-)autonomous
vehicle operation. These last use cases exhibit the most demanding performance
requirements and have especially catalyzed the interest of 3GPP, which further
classified them into four groups, as summarized in Table 1 [1]. Vehicles platooning
dynamically forming a group of vehicles travelling together at short inter-vehicle
distances. Advanced driving enabling vehicles to share local sensor data and driving
intentions with vehicles in proximity, thus coordinating trajectories and maneuvers.
Extended sensors for exchanging raw/processed sensor data or live video among
VUEs, RSUs, VRUs and V2X ASs. Remote driving allowing a remote driver or a
cloud application to tele-operate a (private or public) vehicle; this is useful for those
passengers who cannot drive themselves (e.g., impaired people) or when the vehicle
is located in dangerous or uncomfortable environments (e.g., earthquake-affected
regions, road construction work zones, snow ploughing areas).
The demands of such use cases can hardly be supported by current Radio Access
Technologies (RATs), neither IEEE 802.11 variants nor LTE and C-V2X Releases
14 and 15. This observation motivated the R&D community to explore more per-
forming solutions, entailing not only improvements of the air interface but a more
comprehensive end-to-end approach as provided by 5G.
3 5G for V2X communications
5G systems span communication, networking and computing capabilities, both in

the radio access network (RAN) and in the core network (CN) segments. In the fol-
lowing, the major V2X-related 5G areas of research and enhancement will be shortly
presented, along with solutions falling in the artificial intelligence and security fields
that complete the multidisciplinary 5G-V2X picture.
3.1 Radio access technologies and components
New Radio (NR). Besides further enhancing the PC5 and LTE-Uu interfaces, 3GPP
has launched the NR standardization activity for the first phase 5G system in Release
15, and is ready to enhance C-V2X in several ways under the 5G NR Release 16. NR
will encompass flexible numerologies and agile frame structure, high frequencies,
new multiple access techniques that well answer the quest for high capacity, massive
connectivity, ultra-low latency and high reliability of autonomous driving use cases.
• Millimeter Wave (mmWave) communications ensure a large bandwidth and high
throughput, which can be particularly appealing for: (i) V2V communications
between very close vehicles, e.g., to support cooperative sensing in a high-density
platoon, and (ii) V2I communications for bulk data transfer (e.g., for object
detection and recognition, real-time high-definition maps) to/from an RSU in a
short time frame. The harsh propagation environment may however hamper such
benefits. Challenges arise, for example, due to the overhead for the beam training
under high mobility and the blockage effect by e.g., pedestrian bodies [15].
• Non-Orthogonal Multiple Access (NOMA) allows multiple users to share the
same time/frequency resource by either power-domain or code-domain multiplex-
ing. It provides V2X communications with a new dimension, namely interference
cancellation, therefore improving spectrum efficiency and reducing latency in
dense moving environments (e.g., 2000-4000 vehicles/km2 ). Naively applying
mature OMA-based resource allocation strategies to NOMA is insufficient, e.g.,
scheduling, power control, channel state information (CSI) signaling need to be
rethought in depth [16].
Multi-RATs. 5G will be deployed as a mashup of existing and novel 3GPP (4G
LTE, 5G NR) and non-3GPP (e.g., IEEE 802.11) RATs. In the V2X context, the usage
of multiple RATs may boost V2I/V2N network throughput and capacity, make more
efficient the behaviour of a technology (e.g., mmWave beamforming can be aided by
802.11 messages [15]), or provide redundant connectivity to improve performance
of e.g., the remote driving use cases. The design of a 5G multi-RAT framework is
still an open issue, which needs proper interfaces towards the application layer and
advanced orchestration functionalities for harmonized traffic scheduling and flow
management.
802.11 evolution. On a parallel research lane, probably awaken by the rapid C-
V2X evolution, recently the IEEE 802.11 has started the Next Generation V2X study
group. This group is considering mature physical layer technologies, such as Low
Density Parity Check, Space-time block coding, to have an impact on the evolution
of 802.11 for higher throughput V2X applications, better reliability/efficiency, and
extended range [18].
Transceiver design. Advanced transceivers can be used on board, since vehicles
are not limited by small form factors, processing and power consumption issues.
• Antenna design. 5G will rely on massive Multiple Input Multiple Output
(MIMO), among other techniques, to improve system capacity. The high ve-
hicle speed may however hamper massive MIMO operation due to outdated CSI.
Multi-antenna algorithms need to be designed that are robust against imperfect
CSI, as well as advanced receivers that take advantage of vehicle characteristics.
Even though the form factor of vehicles accommodates a potentially high num-
ber of antennas, their integration in practice might entail new approaches for the
vehicle architecture [26].
• In-band Full Duplex (FD). It is another disruptive technology that improves the
spectral efficiency, theoretically doubling the achievable throughput by simulta-
neous transmission and reception over the same frequency band. Research in this
area has been pushed by recent advancements in self-interference cancellation
(SIC) and/or mitigation techniques. The main barriers for the take-off of FD de-
ployment are the harsh and fast time-varying V2X propagation environment that
complicates SIC procedures, and the need to entirely revisit the MAC design. The
potential of FD techniques has been disclosed for 802.11-based V2X communica-
tions in [11], but similar considerations may apply for 5G V2X communications.
Positioning. Satellite-based positioning systems are unable to provide sufficiently
accurate position information – especially the relative positioning – to critical V2X
applications (e.g., VRU detection, platooning, autonomous driving, self-parking) and
in certain challenging but common environments like urban canyons and tunnels.
Highly-accurate (sub-meter) positioning can be achieved by combining traditional
satellite systems with on-board sensing and infrastructure-based wireless communi-
cation technologies (e.g., 802.11, LTE) and 5G radio-assisted techniques [27].
3.2 Network softwarization technologies
Network softwarization technologies have the potential of efficiently handling het-

erogeneous resources spanning network and cloud domains, and easily and flexibly
deploying services, with significant CapEx and OpEx reduction.
Mobile Edge Computing (MEC). The mainstream and short-term impact of
softwarization for V2X will be at the network edge, where cloud-like resources (i.e.,
computing and storage) can be accessed with a low-latency. The edge is close to where
data are being generated (by on-board and roadside sensors and cameras), and they
are also likely consumed after being processed by the vehicles themselves. The added
value of MEC for V2X has been recently recognized by the 5GAA [6] and by the
European Telecommunications Standards Institute (ETSI) [17]. MEC is particularly
suited for building real-time situational awareness and high-definition local maps,
achieved through real-time data analysis and fusion from multiple available sources
(e.g., vehicles, RSUs). The V2X AS function can be hosted at the edge, in proximity
to vehicles, so that multicast data dissemination as well as mobility management
procedures can be orchestrated with a lower latency [12]. Compared to other vertical
markets, the automotive domain adds a layer of complexity to the MEC design, in
that it entails the migration of services from one edge server to another in response
to vehicles’ mobility. Vehicles can experience performance degradation and ongoing
service interruption during the migration. Seamless service migration between edge
servers is further challenged when occurring across different MNOs, e.g., in a case
of a country border crossing.
Network slicing. Network slicing supports vertical markets with diverse require-
ments on top of the same physical infrastructure, through a flexible usage and
configuration of network functions and parameters. This is viable by drawing on
Software-defined Networking (SDN) and Network Function Virtualization (NFV).
They, respectively, allow to properly steer network traffic with performance guaran-
tees along virtualized network function instances customized for specific use cases.
The potential of the network slicing concept for V2X has been early unleashed in
[12] and further elaborated by 3GPP [2]. There, the need for a dedicated V2X slice
has been argued due to the unique V2X use cases’ features, by recognizing the poor
fitting of reference slices for other traffic types, such as enhanced Mobile Broadband
(eMBB), massive machine type communications (mMTC), ultra-reliable and ultra-
low latency communications (URLLC). By flexibly orchestrating multi-access and
edge-dominated 5G network infrastructures, dedicated network slices for V2X safety
applications can be prioritized over other network traffic. Intra- and inter-MNO mo-
bility entailing, respectively, quick slice resources reconfiguration and inter-operator
slice orchestration highly challenge the design of V2X network slicing.
3.3 Data-driven solutions
With high-performance computing and storage facilities and multiple advanced sen-
sors, vehicles as well as roadside infrastructure will produce, exchange, process and
consume massive amounts of information-rich data used to make driving safer, more
efficient and comfortable. New solutions are needed to facilitate the analysis and
distribution of such data.
Machine learning (ML). As a major branch of artificial intelligence, ML devel-
ops efficient methods to analyze huge (big) amount of data by extracting information
and knowledge which allow more informed and cognitive decisions. The role of ML
in the V2X context will be manifold, with ML methods such as reinforcement learn-
ing, deep learning, deep reinforcement learning and unsupervised learning taking
the main role. By means of ML vehicles can identify crucial data to be primarily
exchanged for cooperative driving purposes with efficient network resources uti-
lization; making ML-assisted decisions also improves radio resource management
algorithms [25]. ML techniques can enhance autonomous driving operations in pro-
ducing accurate models of the surrounding environment from the data generated by
cooperative sensing, and in deriving effective manoeuvring strategies accordingly
[24]. Applying existing ML methods to V2X raises challenges, which are mainly re-
lated to the distributed nature of data produced by multiple sources and the vehicular
dynamics hard to be predicted. Theses issues pave the way to distributed learning
methods. The decision about where to run ML algorithms (i.e., on board, at the edge,
in the cloud) is also another open issue to be addressed.
Information-centric networking (ICN). V2X communications privilege the in-
formation content (e.g., accident notification; stationary vehicle warning; road works
warning) rather than the identity of the communication endpoints. Notably, such
identities are either not known in advance or they may change with time, e.g., the
edge server on the other side of a V2N link may change when a vehicle moves.
In addition, the same information can be of interest for multiple recipients on the
road, so that V2X interactions will be dominated by one-to-many communications,
which are notoriously poorly supported by IP-based solutions. ICN natively supports
name-based data retrieval and dissemination, and multicast delivery, which perfectly
fit the above mentioned V2X characteristics. Furthermore, in-network caching can
cope with intermittent connectivity. These features make ICN a candidate solution
for V2X data delivery [9]. Despite such potential and the fact that 3GPP is open to
networking solutions that may replace and/or work alongside IP, the proper way ICN
could be enclosed in the 5G networking still needs to be adequately investigated.
3.4 Security
The 5G-envisioned ubiquitous vehicle connectivity demands strong security and

privacy mechanisms to prevent unauthorized access to vehicles and related personal
data, and to avoid the malfunctioning of mission-critical applications and future
automated driving systems and services [26]. The overhead introduced by the security
features (e.g., certificate revocation), typically addressed at the application layer [2],
should be reduced, and the timeliness of authentication management should be
ensured by accounting for new security implications raising in V2X environments
[22]. Recent works have investigated the role that the blockchain (BC) technology
could have in a V2X environment. In [10], the decentralized validation architecture
of BC, without a trusted third party authority, is used for fast validation of messages
exchanged within a platoon by applying a consensus-rule. Last, new security threats
could specifically arise with network slicing. V2X network slices with different
security assurance requirements may coexist, aiming at ensuring adequate isolation
between them and other reference slices [12].
4 Conclusions
In this chapter, we have analysed the technical issues and opportunities related to
the V2X support in 5G systems. The analysis clearly shows that the automotive
vertical highly challenges 5G entailing the synergies of many stakeholders to meet
the stringent needs of V2X use cases. For the full take off of connected and automated
vehicles, non-technical barriers must be also removed. They are related to legal and
liability issues (who will be liable when a driverless car is involved in a collision?
the car’s occupants, the auto maker or the software company?); ethics (should an
automated car decide for the driver on ethical questions of life and death?); the need
to foster end-user acceptance (will we accept it when machines make mistakes, even
if they make far fewer mistakes than humans?); and to build solid business models.
References
1. 3GPP TR 22.186 v16.0.0. Technical specification group services and system aspects. En-
hancement of 3GPP support for V2X scenarios, Rel. 15. Sept. 2018.
2. 3GPP TR 23.786 v0.8.0. Technical specification group services and system aspects; Study on
architecture enhancements for EPS and 5G system to support advanced V2X services, Rel.
16, Aug. 2018.
3. 3GPP TR 36.213 v15.3.0, Technical specification Group Radio Access Network. Evolved
Universal Terrestrial Radio Access (E-UTRA); Physical layer procedures, Rel. 15. Oct. 2018.
4. 3GPP TS 23.285 v15.1.0. Technical specification group services and system aspects; Archi-
tecture enhancements for V2X services, Rel. 14. June 2018.
5. 5G Automotive Association (5GAA). http://5gaa.org/
6. 5G Automotive Association (5GAA) - Toward fully connected vehicles: Edge computing for
advanced automotive communications. December 2017.
7. 5G Communication Automotive Research and Innovation (5GCAR) H2020 project.
http://5gcar.eu/
8. 5G for connected and automated road mobility in the european union (5G-CARMEN) H2020
project. http://5gcarmen.eu/
9. Amadeo A, Campolo C, Molinaro A (2016) Information-centric networking for connected
vehicles: a survey and future perspectives. IEEE Communications Magazine, 54(2): 98–104.
10. Calvo J. A. L, Mathar R (2018) Secure blockchain-based communication scheme for connected
vehicles. In IEEE European Conference on Networks and Communications (EuCNC).
11. Campolo C, Molinaro A, Berthet A.O, Vinel A (2017) Full-duplex radios for vehicular com-
munications. IEEE Comm. Mag., 55(6): 182–189.
12. Campolo C, Molinaro A, Iera A, Menichella F (2017) 5G network slicing for vehicle-to-
everything services. IEEE Wireless Communications, 24(6): 38–45.
13. Campolo C, Molinaro A, Scopigno R (2015) From today’s vanets to tomorrow’s planning and
the bets for the day after. Vehicular Communications, 2(3): 158–171.
14. Campolo C, Molinaro A, Scopigno R (2015) Vehicular ad hoc networks: Standards, Solutions,
and Research. Springer-Verlag.
15. Choi J, et al. (2016) Millimeter-wave vehicular communication to support massive automotive
sensing. IEEE Communications Magazine 54(12): 160–167.
16. Di B, Song L, Li Y, Han Z (2017) V2X meets NOMA: Non-orthogonal multiple access for
5G-enabled vehicular networks. IEEE Wireless Comm., 24(6):14–21.
17. ETSI - Multi-access Edge Computing (MEC); Study on MEC Support for V2X Use Cases.
Sept. 2018.
18. IEEE 802.11 Next Generation V2XSG, http://www.ieee802.org/11/reports/ngvsg_update.htm.
19. MacHardy Z, Khan A, Obana K, Iwashina S (2018) V2X access technologies: Regulation,
research, and remaining challenges. IEEE Communications Surveys & Tutorials, 20(3): 1858–
1877.
20. Masini B, Bazzi A, Zanella A (2018) A survey on the roadmap to mandate on board connec-
tivity and enable V2V-based vehicular sensor networks. Sensors, 18(7).
21. Molina-Masegosa R, Gozalvez J (2017) LTE-V for sidelink 5G V2X vehicular communi-
cations: a new 5G technology for short-range vehicle-to-everything communications. IEEE
Vehicular Technology Magazine, 12(4): 30–39.
22. Muhammad M, Safdar G. A (2018) Survey on existing authentication issues for cellular-
assisted V2X communication. Vehicular Communications, 12: 50–65.
23. Nguyen T.V, et al. (2017) A comparison of cellular vehicle-to-everything and dedicated short
range communication. In IEEE Vehicular Networking Conference (VNC).
24. Sallab A. E, Abdou M, Perot E, Yogamani S (2017) Deep reinforcement learning framework
for autonomous driving. Electronic Imaging, 2017(19): 70–76.
25. Ye H, et al. (2018) Machine learning for vehicular networks: Recent advances and application
examples. IEEE Vehicular Technology Magazine, 13(2): 94–101.
26. White Paper on Automotive Vertical sector, 5G automotive vision, 5G-PPP. October 2015.
27. Wymeersch H, et al. (2017) 5G mmWave positioning for vehicular networks. IEEE Wireless
Communications, 24(6): 80–86.
Rapid Deployment of 5G Services Using Drones
and other Manned and Unmanned Aerial
Vehicles
Riccardo Bassoli, Fabrizio Granelli
Abstract 5G Networks are expected to introduce several breakthroughs and a big step
forward towards a flexible and effective mobile network. An interesting requirement
for the 5th generation of cellular networks is indeed the possibility to deploy a
network in a very short time frame, indicatively 90 minutes. The purpose of this
unprecedented design goal is to address scenarios of dynamic coverage requirements,
especially targeted at unexpected or emergency situations.
In this framework, a research proposal is being developed at the CNIT Research
Unit at the University of Trento in collaboration with Technion in Israel to define
and prototype a suitable architecture to provide on-demand 5G coverage for border
monitoring and disaster scenarios.
1 Introduction
5G is expected to provide a big step forward in enabling fast deployment of networks,

shifting the time scale from days to hours or less. Indeed, 3GPP requirements for 5G
cellular networks propose a nominal deployment time of 90 minutes. This feature
will enable 5G to offer connectivity and services in novel relevant scenarios, such as
border monitoring and surveillance and disaster scenarios.
In this framework, several works in the literature outlined the possibility to inte-
grate moving BSs within the 5G infrastructure, using Manned or Unmanned Vehicles
- in several cases Aerial Vehicles. Aerial vehicles provide several advantages over
land vehicles due to their agility, freedom of operation and potential coverage, at the
cost of limited lifetime and range of operation.
Riccardo Bassoli
DISI - University of Trento, Via Sommarive 9, Trento, ITALY, e-mail: riccardo.bassoli@unitn.it
Fabrizio Granelli
Consorzio Nazionale Interuniversitario per le Telecomunicazioni - Sede di Trento, Via Sommarive
9, Trento, ITALY, e-mail: fabrizio.granelli@unitn.it
219
220 Riccardo Bassoli, Fabrizio Granelli
Despite the potential benefits of unmanned aerial vehicles in applications like

disaster recovery, environmental monitoring, flood area detection, and aerial surveil-
lance of public areas, there are still several open issues to be addressed. These are
mainly related to their effective usage for data/information collection, communi-
cation and processing. Moreover, regulation about data access requires an efficient
selection of authorized personnel to manage sensitive information. Finally, a network
of unmanned vehicles needs to adapt to unpredictable events, attacks and network
dangerous states to guarantee optimal quality of monitoring experience: eventually,
the network should be capable not only to detect but also to prevent dangerous
network situations.
In this context, the research activities of the Dynamic Architecture based on UAVs
Monitoring for border Security and Safety (DAVOSS) project (nato-davoss.org),
funded by the NATO in the framework of the Science for Peace and Security Pro-
gramme (funding period: 2018-2021), aim at advancing the current monitoring
networks based on unmanned aerial vehicles to help to overcome some of their tech-
nological limitations, and focusing on system reliability. The project will study and
design a virtualized cloud-based architecture to enhance capabilities of current bor-
der surveillance and counter-terrorist operational networks based on sensors, cameras
and unmanned aerial vehicles. The DAVOSS solution will consider different kind
of environments. Moreover, given its dynamic network structure and adaptability,
it will provide higher security against physical attacks and natural catastrophes.
The centralized structure of the architecture will allow for easier implementation of
traffic measurement and anomaly detection processes, even in case of disaster fore-
cast: its dynamic reconfiguration will optimize network performance, information
management and processing, by ensuring optimal coverage to sensors and moni-
toring peripherals. Finally, the architecture will define and develop an appropriate
wide-range connectivity functionality to provide the most suitable communication
paradigm for connecting with the remote control center, via 4G/5G cellular back-
haul, through the intervention of an Ultra Light aerial Vehicle (UVL) - such a small
airplane, nano-satellite or blimp, multi-hop wireless mesh networking or, in a possi-
ble future scenario, the usage of satellites. In particular, lightweight flying platforms
such as ULV or small satellites will represent a viable alternative to terrestrial back-
haul in terms of easy and low cost deployment and robustness against attack and
environmental disruption. The proposed solution will also prevent information leak-
age, since no sensitive military/security data will be processed at unsecure network
entities. It will also possible to easily monitor the effectiveness and the efficiency of
network updates since they will be performed in a centralised manner.
This chapter will present the architecture and current status of the DAVOSS 5G
Testbed in Trento.
Rapid Deployment of 5G Services Using Drones 221
2 Related Works
The usage of advanced communication and networking infrastructures for border

monitoring and emergency scenario is commonly found in the literature. In this
section we review some of the most interesting approaches, focusing on the cases
including WAN connectivity in addition to local coverage.
In [1], a solution based on the integration of Ku-band radar systems installed
on UAVs and GNSS localization is proposed for patrolling of sea borders in the
Mediterranean area. In [2], the use of aquatic drones is considered for marine safety.
In these works, the focus is on the optimal sensor deployment and on the best routing
approach, exploiting state-of-the-art technology and standard network configuration.
Other papers, like e.g. [3] and [4] consider the use of UAVs in combination with
ground sensors in order to foster and optimize the border monitoring and minimizing
the false alarms. However, such approaches are not based on an effective integration
of the different network infrastructures involved and still rely on human operators’
intervention to work.
In [5], Kim, Mokdad and Ben-Othman analyze the design of UAV-based surveil-
lance networks in two different scenarios: the smart city and the extensive ocean.
Differentiated UAV typologies and network configurations are proposed in [5] for
the two scenarios, evidencing a substantial weakness of UAV-based monitoring in
terms of lack of adaptation to potential modifications of the test field.
The DAVOSS approach represents a step forward with respect to the current
state-of-the-art about the use of avionic networks for environmental and border
monitoring. Indeed, the flexibility and reconfigurability introduced by the DAVOSS
network architecture in terms virtualization and softwarization is expected to provide
a viable yet effective solution to adapt to dynamic changes of the application scenario.
3 The DAVOSS Network Concept
The DAVOSS project proposes to define a virtualized cloud-based architecture based

on different types of manned and unmanned aerial vehicles, to enhance the capa-
bilities and the reliability of current border surveillance and disaster management
systems.
3.1 Global Architectural Overview
Figure 1 depicts the proposed DAVOSS architecture. The system can be divided into
four main layers:
• Layer 1 consists of the ground-level sensors and peripherals, which are devoted
to different kind of sensing procedures according to the application scenario and
the environment.
• Layer 2 represents the fleet of UAVs equipped with a camera and hardware for
data transmission/reception. The UAVs provide network connectivity (by acting as
mobile gateways) and further monitoring functionalities both in case of disasters
and border security/terrorist attacks.
• Layer 3 provides network and resources virtualisation, and manages virtual net-
work function assignment and slicing. This layer will implement a Software
Defined Networking approach to control the connectivity and performance of the
underlying mobile nodes (e.g. the UAVs), and well as Network Function Virtual-
ization to assign or re-locate relevant processing and security functionalities.
• Layer 4 (Wide-Area connectivity) is responsible to collect information from UAVs
and to transmit it securely to the cloud servers located at the remote control center.
Different solutions for communication with cloud servers will be analyzed, tested
and experimented, including direct usage or mesh-based solutions for efficient
usage of the existing 4G/5G cellular infrastructure as backhaul, usage of a manned
ULV to collect data by the virtualized network of UAVs and sensors in a delay-
tolerant paradigm, usage of satellite communication (CubeSat scenario). This
layer will be the key to guarantee coverage, security, availability and reliability,
in case of both disasters and terrorists threats.
The project testbed will implement a subset of the solutions at Layer 4, considered
to be the best ones, but will also investigate future extension of the architecture
through CubeSat or other advanced solutions.
3.2 Sensor Network Deployment Solutions
The project will study optimal Sensor Network Deployment solutions, mainly based
on Low Power Wide Area Network (LPWAN) wireless telecommunication technolo-
gies [6]. The basic characteristics of this technology are: (a) ability to inter-connect
battery-powered end-devices over long ranges, (b) the end-devices must operate at
low power, (c) downlink and uplink traffic is at low bit rate (0.3 kbit/s to 200 kbit/s)
per frequency channel, (d) the frequencies used are licensed or unlicensed , (e) pro-
prietary or open standard protocols are used. The following technologies are the most
popular: Sigfox, LoRa, NB-IOT (Narrowband IOT) , LTE-M. We examined closely
Sigfox and LoRaWAN and found the main characteristics as described in Tab. 1.
Based on the above considerations, DAVOSS focuses on LoRa technology and to
use LoRaWAN as the MAC protocol for the Network Deployment solution.
The technical specification of LoRa/LoRaWAN is:
• LoRa ISM Band : 868MHz - 900MHz (EU) , 902MHz - 928MHz (US);
• Ranges: 5 km (Urban) - 15 km (LoS);
• Security: Authentication and Encryption AES-128;
Fig. 1 Structure of the DAVOSS proposed system.
Table 1 Sigfox and LoRa standard comparison

Sigfox LoRa
Narrowband (or Wide band (125Khz or more)
ultra-narrowband) technology Spread-Spectrum technology
Uses a standard radio Uses on frequency-modulated chirp
transmission method (BPSK) Wide band (125Khz or more)
Requires an inexpensive End node radio,
Both the End node and the Gateway are
but expensive HW at the Gateway relatively inexpensive
Looks at a wider
amount of spectrum than SigFox
Uplink quality: good,
so can get more Interference.
Downlink quality: Limited
The larger receiver frequency
bandwidth is mitigated by the coding gains
Anyone can join the LoRa Alliance. LoRa Gateway
spec is open. LoRaWAN which is the MAC
Technology and protocols from the end
protocol above LoRa is an open standard
node to the server are not open.
developed by committee.
Network management spec is open.
• Data Rates: 0.3Kbps – 50Kbps.

The LoRaWAN specification version 1.1 defines 3 device classes:
• Class A devices have the lowest power consumption by opening two short receive
windows after transmission.
• Class B devices extend Class A by adding slotted communication.
• Class C devices extend Class A by keeping the receive windows open unless they
are transmitting.
3.3 Software Defined Networking and Network Function Virtualization
Layer 3 will aim at maximising system automation and autonomy, by providing

centralised configuration, quick, reliable and secure access to information, and en-
capsulating information at different user’s levels – thanks to key 5G technologies of
SDN and slicing.
Software Defined Networking represents an emerging paradigm which enables
the separation of control functionalities from traditional Internet routers in order to
transform them into dumb "Switches" controlled by a central entity (namely: the
SDN controller) [7]. SDN demonstrated its effectiveness in improving the control
and programmability of the current packet networks. Indeed, the SDN controller,
having a central and global vision of the whole network, is capable of optimizing
the performance, managing in effective manner the various traffic flows and finally
guaranteeing a satisfactory Quality of Service (QoS) to the users.
Virtualisation technologies will efficiently handle different kinds of traffic, with

different priority. DAVOSS will provide the necessary centralised-cloud commu-
nications system. When available, other networks such as cellular networks or the
Internet will also provide the required connectivity and infrastructure. Based on
these communication technologies, DAVOSS also aims to exploit adaptive slicing.
That will be used to bring rich computational and network resources to authorized
UAVs. Authorized end users will have more information by increasing the number
of information gathering nodes, real-time availability, and interoperability among
systems: that is made possible by deploying dynamic slicing. With the amount and
quality of information available in real-time, action will be immediately steered to
the location of interest. A centralized analysis of network status, and of data about
border surveillance will prevent network monitoring to fail because of attacks and
lack of resources. Last, but not the least, UAVs, cameras and sensors with dynamic
virtualisation and slicing will significantly reduce intensive human interaction and
control.
DAVOSS virtualization approach with adaptive slicing represents a novelty with
respect to the state-of-the-art. A very recent work [8] considered the use of SDN and
virtualization in UAV networks. However, the SDN architecture of [8] is targeted
at managing multi-path routing only, by searching for the best available path. In
DAVOSS approach, SDN and virtualization are regarded as complementary tools

capable of dynamically and adaptively manage the overall link and processing re-
sources involved in the border patrolling tasks, while ensuring the maximum possible
reliability and network lifetime.
3.4 Satellite- or blimp-based Backhauling
The backhaul plays a key role in the DAVOSS network architecture. Indeed, the
information acquired by the ground sensors and processed by the drone layer should
be forwarded in real time to remote control stations, that may be considerably far
from the monitored area. Moreover, the DAVOSS system considers scenarios where
the terrestrial network connection may not be available (e.g. desert/mountain areas
or open sea). For this reason, effective and reliable backhauling pays a key role in
the architecture.
In this framework, the use of satellite links for long-range data transmission
in emergency recovery and public safety applications is regarded as a resilient
solution, whose deployment costs are limited and convenient [9]. Geostationary
(GEO) satellites present very favorable coverage and availability, but, as drawback,
they are characterized by high latency due to the very long distance from Earth. Low-
Earth-Orbit (LEO) satellites placed at orbital heights of 500-700 Km offer reduced
coverage with respect to the GEO counterparts, but also acceptable latency.
In the framework of DAVOSS research, different alternatives will be studied,
involving the usage of blimps, Ultra Light Aerial vehicles or small satellites.
One of the novel solutions for long-distance backhaul will be based on the use
of the CubeSat picosatellites. Nowadays, CubeSats are raising a lot of interest in
the aerospace research community thanks to the reduced development and launch
costs. Despite to their small amount of available volume, CubeSat missions have
been proven to be very effective in high added-value applications like scientific
data gathering, educational purposes and small-scale industrial equipment testing
[14]. The on-board processing capabilities of CubeSats are not so limited as one can
expect. Indeed, the use of dedicated processors, based e.g. on FPGA technology [15],
allows to perform on-board image processing [14] [15] with fully-affordable power
consumption. As far as communication aspects are considered, considerable research
efforts have been done in order to overcome the bottleneck of low-rate standard radio
interfaces, like e.g. AX-25 or similar variants [16], capable of providing small
throughput of the order of 9.6 Kb/s. In [16], an X-band CubeSat communication
system, compatible with the NASA Near Earth Network,offering a downlink data
rate of 12.5 Mb/s has been implemented and tested. In [17], a prototype of 2.4 GHz
High-Data Rate (HDR) radio for CubeSat has been implemented, able at supporting
a topic data rate of 60 Mb/s. We believe that these last numbers and consideration
can fully justify the CubeSat solution for DAVOSS long-range communication, thus
solving the tradeoff between costs and coverage (the footprint diameter of a single
CubeSat is well enough for DAVOSS purposes).
Nevertheless, CubeSats will be considered in the design and simulation phases

of the project, and for future implementations of the project. DAVOSS testbed and
proof-of-concept will be based on locally available alternatives for backhauling,
including LTE/5G, Ultra Light Aerial Vehicle and Helium-based Blimps.
4 DAVOSS Network Design Preliminary Assessments
This section presents some preliminary results of the model used for analyzing the
DAVOSS networking infrastructure, focusing on the key aspects of energy consump-
tion and virtualization of BBUs. In particular, the analysis is focused on comparing
the different alternatives of re-location of BBUs: at UAVs, at geostationary satellites,
or at CubeSats. The model is based on stochastic geometry, in order to calculate the
variation of the average number of v-BBUs and the impact of virtualisation on the
power consumption of the system.
In order to provide a realistic data of BBUs, the technical specifications of the
Ericsson-Baseband-5212-5216 [10] are used. However, the generality of the model
allows the correct use of any BBUs’ data sheet. The average traffic provided con-
stantly by peripheral sensors is set to 500 kb/s.
The deployment of virtualisation allows proper optimization in a dynamic net-
working scenario, in which only v-BBUs are considered, which are activated accord-
ing to network and traffic requirements. This does not happen in current monitoring
networks based on 4G/LTE, where each active mobile base station must always host
an active BBU or the split between BBUs and RRHs is performed a-priori.
Given λbs = 30 AP/km2 and λs = 900 peripherals/km2 , this means that a 4G/LTE-
based monitoring network maintains active 30 BBUs/km2 . The energy consumption
of a BBU can be estimated to be 3 W for pico cells mobile BSs [11].
Fig. 2 shows the Voronoi tessellation of a unit of area to depict the properties
related to coverage.
Given these premises, the value of peripherals that a mobile BS has to serve, with
higher probability, is 22.
The resulting relationship between weight of the drone (mobile BS) and the power
consumption is depicted in Fig. 3. In particular, the gain is calculated in respect of
mobile BSs, which carry BBU weighting less than 4 kg.
Let’s consider the possibility to re-locate the BBUs within geostationary satellites.
For the considered border area, it has to handle λbs Au , where Au is the unitary area.
That means 41340 mobile BSs. Given the limited capacity of a v-BBU [10] at the
geostationary and CubeSat satellites, the datacentre processors will serve mobile
BSs according to a queueing model. Detailed analysis of this aspect of the scenario
will be considered in future works.
Regarding latency, the total delay of the two approaches can be modelled as
ttotnoV = t prop + tBBU proc + tRRH (1)

Fig. 2 Voronoi tessellation, which provides a snapshot of the coverage of a unit of area. The red
stars are the mobile BSs while the blue dots are the peripherals.
Fig. 3 Power gain at the mobile BSs (drones) when BBU is virtualised. Obviously, by increasing
the load of the drone, the impact of the weight of the BBU decreases.
and
ttotV = t prop + tCloudproc + tRRH + tback (2)
where ttotnoV and ttotV are the total latency without virtualisation and with BBU
virtualisation respectively. In particular, t prop is the propagation delay, tBBU proc is
the processing time of a physical BBU, tCloudproc is the processing time in the cloud
(i.e. the satellite) of v-BBUs, tRRH [13] is the remote radio head (RRH) delay and
tback is the backhaul latency.
By considering the values of latency for Legacy long term evolution (LTE)
uplink in [12], equation (1) and equation (2) becomes respectively ttotnoV = x + 2.5,
ttotV = x + 121.5 (GEO satellite) and x + 4 ≤ ttotV ≤ x + 4.66 (CubeSat). The
latencies of these formulas are measured in ms.
Fig. 4 Behaviour of total latency functions depending on the increase in processing time at physical
BBUs or vBBUs at satellites.
As clearly appears by Fig. 4, the trade-off between reduction in energy consump-

tion and latency becomes significant when satellites are involved in the Cloud RAN
realisation. Furthermore, it also becomes clear that the choice of CubeSats is funda-
mental to have reasonable response time in case of data transmissions whose quality
is hardly affected by latency. In that sense, a possible vision of DAVOSS to choose
ultralight aerial vehicles or remotely operated blimps as an alternative to satellites
to host cloud computing shows its importance.
On the other hand, the deployment of physical BBUs at the UAVs is an optimal
choice in terms of latency but it increases a lot the energy cost at the drones. That
would probably involve, given that a fleet of UAVs have very short flight time
and require very frequent charging time, an extremely dynamic and challenging
networking environment.
For more information on the model, please refer to [18].
Ongoing work is focused on analyzing the different solutions of backhauling, that
will be reported on the project website: nato-davoss.org
5 Conclusions
5G design will include the possibility of fast deployment of the networking infras-
tructure. This would represent a great advantage in application scenarios, such as
border monitoring and disaster situations.
This chapter analyzed an ongoing project activity at the Research Unit of Trento
in order to design, implement and demonstrate a 5G architecture able to provide
connectivity and advanced services using network virtualization and UAVs/aerial
communication platforms. The results of this project will provide useful sugges-
tions about the possibility of generating and managing reliable yet fast connectivity
solutions using drones and 5G technology.
For additional information and updates related to the project, please visit the
DAVOSS project website: nato-davoss.org
Acknowledgements This work has been partially funded by NATO Science for Peace and Security
(SPS) Programme, in the framework of the project SPS G5428 "Dynamic Architecture based on
UAVs Monitoring for Border Security and Safety".
References
1. D. Tarchi and G. Guglieri and M. Vespe and C. Gioia and F. Sermi and V. Kyovtorov, "Search
and Rescue: Surveillance support from RPAs radar". 2017 European Navigation Conference
(ENC), 256–264 (2017).
2. D. Moura and L. Guardalben and M. Luis and S. Sargento, "A Drone-Quality Delay Tolerant
Routing Approach for Aquatic Drones Scenarios". 2017 IEEE Globecom Workshops (GC
Wkshps), 1–7 (2017).
3. D. Bein and W. Bein and A. Karki and B. B. Madan, "Optimizing Border Patrol Opera-
tions Using Unmanned Aerial Vehicles". 2015 12th International Conference on Information
Technology - New Generations, 479–484 (2015).
4. L. L. Coulter and D. A. Stow and Y. H. Tsai and C. M. Chavis and R. W. McCreight and C. D.
Lippitt and G. W. Fraley, "A new paradigm for persistent wide area surveillance". 2012 IEEE
Conference on Technologies for Homeland Security (HST), 51–60 (2012).
5. H. Kim and L. Mokdad and J. Ben-Othman, "Designing UAV Surveillance Frameworks
for Smart City and Extensive Ocean with Differential Perspectives". IEEE Communications
Magazine, Vol. 56, No. 4, pp. 98–104 (2018).
6. U. Raza and P. Kulkarni and M. Sooriyabandara, "Low Power Wide Area Networks: An
Overview". IEEE Communications Surveys Tutorials, Vol. 19, No. 2, pp. 855–873 (2017).
7. D. B. Rawat and S. R. Reddy, "Software Defined Networking Architecture, Security and
Energy Efficiency: A Survey". IEEE Communications Surveys Tutorials, Vol. 19, No. 1, pp.
325–346 (2017).
8. G. Secinti and P. B. Darian and B. Canberk and K. R. Chowdhury, "SDNs in the Sky: Robust
End-to-End Connectivity for Aerial Vehicular Networks". IEEE Communications Magazine,
Vol. 56, No. 1, pp. 16–21 (2018).
9. M. Panizza and C. Sacchi and J. Varela-Miguez and S. Morosi and L. Vettori and S. Digenti
and E. Falletti, "Feasibility study of a SDR-based reconfigurable terminal for emergency
applications". 2011 Aerospace Conference, 1–18 (2011).
10. Ericsson, Baseband Description – Baseband 5216, Baseband 5212, 2016,
http://www.1com.net/wp-content/uploads/2017/03/Ericsson-Baseband-5212-5216-
datasheet-specs.pdf
11. G. Auer and V. Giannini and I. Godor and P. Skillermark and M. Olsson and M. A. Imran
and D. Sabella and M. J. Gonzalez and C. Desset and O. Blume, "Cellular Energy Efficiency
Evaluation Framework". IEEE 73rd Vehicular Technology Conference (VTC Spring), 1–6
(2011).
12. S. Nagata and L. H. Wang and K. Takeda, "Industry Perspectives". IEEE Wireless Communi-
cations, Vol. 24, No. 3, pp. 2–4 (2017).
13. L. Zhang and A. U. Quddus and E. Katranaras and D. Wübben and Y. Qi and R. Tafazolli,
"Performance Analysis and Optimal Cooperative Cluster Size for Randomly Distributed Small
Cells Under Cloud RAN". IEEE Access, Vol. 4, pp. 1925–1939 (2016).
14. E. Baceski and S. Gokcebag and A. Erdem and C. G. Erbay and M. Akyol and K. Arslankoz
and I. Arslan and M. A. Agca and Y. B. Aydin and A. R. Aslan and O. Ceylan, "HAVELSAT:
A software defined radio experimentation CubeSat". 2015 7th International Conference on
Recent Advances in Space Technologies (RAST), 831–834 (2015).
15. D. L. Bekker and T. A. Werne and T. O. Wilson and P. J. Pingree and K. Dontchev and M.
Heywood and R. Ramos and B. Freyberg and F. Saca and B. Gilchrist and A. Gallimore and J.
Cutler, "A CubeSat design to validate the Virtex-5 FPGA for spaceborne image processing".
2010 IEEE Aerospace Conference, 1–9 (2010).
16. S. E. Palo, "High rate communications systems for CubeSats". 2015 IEEE MTT-S International
Microwave Symposium, 1–4 (2015).
17. B. Butters and R. Raad, "A 2.4 GHz High Data Rate radio for pico-satellites". 2014 8th
International Conference on Telecommunication Systems Services and Applications (TSSA),
1–6 (2014).
18. R. Bassoli, C. Sacchi, F. Granelli, I. Ashkenazi, "A Virtualized Border Control System based
on UAVs: Design and Energy Efficiency Considerations". 2018 IEEE Aerospace Conference
(2018).
5G Planning under EMF Constraints
Angela Sara Cacciapuoti, Luca Chiaraviglio, Gerardo Di Martino and Marco Fiore
Abstract The planning of a 5G cellular network requires the installation of 5G Base

Station (BS) sites and the configuration of each BS. This process takes into account
several constraints, including the ElectroMagnetic Fields (EMF) levels radiated by
the installed BSs. Different countries in the world (including Italy) adopt EMF
constraints which are much stricter than the ones prescribed by international orga-
nization. The resulting planning is far away to be optimal, resulting in an increase
of the CAPital EXpenditures (CAPEX) costs for the operator, and a decrease of the
user Quality of Service (QoS). In this chapter, we shed light on this problem. After
reporting the related work, we discuss the impact of strict EMF constraints on the 5G
planning through a set of representative examples. We then report the outcomes from
two case studies. Results clearly show that a saturation of EMF levels, preventing
the installation of 5G BS sites, is already reached in currently deployed networks.
In addition, we also quantify that the negative impact of the non-optimal planning
on the user QoS. Finally, we discuss the expected impact of the main 5G technology
features on the EMF levels.
Angela Sara Cacciapuoti

Department of Electrical Engineering and Information Technology (DIETI), University of
Naples Federico II, and Consorzio Nazionale Interuniversitario per le Telecomunicazioni, e-mail:
angelasara.cacciapuoti@unina.it
Luca Chiaraviglio
Department of Electronic Engineering, University of Rome Tor Vergata, Rome, Italy, and Con-
sorzio Nazionale Interuniversitario per le Telecomunicazioni, Italy e-mail: luca.chiaraviglio@
uniroma2.it
Gerardo Di Martino
Department of Electrical Engineering and Information Technology (DIETI), University of
Naples Federico II, and Consorzio Nazionale Interuniversitario per le Telecomunicazioni, e-mail:
gerardo.dimartino@unina.it
Marco Fiore
Institute of Electronics, Computer and Telecommunication Engineering, National Research Council
of Italy, Turin, Italy, e-mail: marco.fiore@ieiit.cnr.it
231
232 Angela Sara Cacciapuoti, Luca Chiaraviglio, Gerardo Di Martino and Marco Fiore
1 Introduction
In the next decade, 5G is expected to enable a plethora of new services, with a

potential economic impact of several trillions of dollars [?]. The success will depend
on the capability of the 5G technologies to meet strong performance requirements
such as a dramatic increase of user throughput up to 10 Gbps, or extremely low
order-of-millisecond communication latency.
The combined utilization of technologies, such as, e.g., massive MIMO, adaptive
beamforming, mmWave, softwarization of the network functions, is expected to
guarantee the required increase of the offered capacity, along with the flexibility and
efficiency in the network management. In many countries, and among them in Italy,
the first field trials will be operative by the end of 2018.
In this context, a very delicate and frequently underrated issue is related to the
compliance of the new 5G deployments with regulations on Electromagnetic Field
(EMF) exposure limits [?], which impose limits on the EMF levels resulting from
the composition of the different radiating sources.
The EMF limits set by the International Commission on Non-Ionizing Radiation
Protection (ICNIRP) are widely adopted around the world. However, several other
countries (e.g., China, India, Russia, Switzerland, Italy, Canada and Poland) impose
even more stringent limits. Since 5G will operate on a very wide range of frequencies
(ranging from hundreds of MHz, to provide coverage in rural areas, up to tens
of GHz to support very high data rates) and it will coexist with some pre-5G
technologies, stringent EMF limits are expected to have an especially strong impact
on the planning of forthcoming 5G networks. Indeed, network planning is a complex
problem per se. It aims at minimizing the CAPital EXpenditures (CAPEX) by the
network operator, by jointly (i) selecting proper locations for the sites hosting the
Base Stations (BSs), (ii) dimensioning the radio equipment installed at each BS, and
(iii) fulfilling performance constraints on coverage, offered capacity and Quality of
Service (QoS) perceived by end users. Factoring in both EMF exposure limits and
the specificities of 5G radio technologies significantly complicates the problem.
Stemming from the above, a planning phase neglecting EMF constraints neces-
sarily introduces an increased risk of sub-optimal planning, with an associated need
of a posteriori refinements, impacting both the operator CAPEX and the user QoS
[?]. However, despite its prominent role in the 5G roll-out, the 5G network planning
under EMF constraints is still completely open.
In this chapter, we present a comprehensive analysis of this topic, starting from
the work presented in [?]. Specifically, first, in Section 2 we review the current state
of the art by analyzing different aspects related to EMF exposure levels: health risks,
EMF measurements campaigns, and cellular network planning. In Section 3 we
contextualize the EMF limits in Italy and we discuss their effects on the planning of
5G cellular networks. In support of the presented analysis, in Section 4, we present
two Italian case studies. The first one provides evidences of EMF saturation already
in presence of sole pre-5G technologies. The second case study provides evidences
on the impact of restrictive EMF limits on the planning of a 5G deployment, and
hence on the QoS perceived by the users. In Section 5 we outline the expected impact
5G Planning under EMF Constraints 233
of the main 5G technology features on the EMF exposure levels. Finally, Section 6
draws relevant conclusions.
2 Related Work
We classify the related work in the following categories: i) health risks from EMFs
due to cellular networks, ii) EMF measurements in mobile networks, iii) cellular
planning with EMF constraints.
2.1 Health Risks of EMFs Exposure from Cellular Networks
The EMFs generated by the cellular network may trigger two types of risks associated
with the health: i) heating of the radiated tissues, and ii) non-heating effects of the
exposed body. Focusing on the first effect, there is a large literature clearly showing
the negative effects on the health (we refer to reader to [?] and references therein).
In this context, the ICNIRP has adopted a set of guidelines to adhere in order to
avoid the heating effects [?]. These guidelines include the definition of EMF limits
which differentiate between the workers that perform e.g., maintenance operations
in proximity to a BS site and are aware of the risk, and the general public, which is
not aware of the risk and need to be preserved through strict limits. In particular, the
EMF limits defined by ICNIRP are several times below the critical values triggering
heating effects.
Focusing instead on the non-heating effects, these phenomena may include cancer
diseases appearing in mature life, which may be triggered even if the EMFs absorbed
by the tissues are lower than the ICNIRP limits. In this context, the impact of non-
heating EMF effects on the health is still a matter of research. The Interphone project
[?, ?] started in 2000 as an international effort from different countries in the world
in order to monitor the relantionship between the use of mobile phones and the
appearance of tumors in the brain, the acoustic nerve and the parotid gland. The
outcomes suggested that there is increased risk in the brain tumor. However, as noted
by the World Health Organization (WHO) [?] bias and errors may have impacted
the conclusions of [?]. Therefore, a causal interpretation between the use of mobile
phone and the presence of tumors can not be found.
Recently, Falcioni et al. [?] performed several long-term experiments involving
the exposure of rats to large EMFs generated by GSM equipment. Their results show
a statistical increase in the incidence of heart tumor in rats exposed for a long period
of time to an EMF of 50 [V/m]. In addition, an increase in both the heart Schwann
cells hyperplasia and malignant glial tumors were observed (although not statistically
significant). In this context, the United Station (US) National Toxicology Program
(NTP) has recently concluded another study, which involved the exposure of rats to
the EMF generated by 2G and 3G equipment [?, ?]. Their outcomes show a clear
evidence of heart tumors in the rats (similarly to [?]), while some evidence of tumors
in the brains and in the adrenal glands. It is important to remark that both the studies
[?, ?, ?] have been conducted considering legacy technologies, while the impact of
currently deployed one (i.e., 4G) and future ones (5G) is still an open issue.
In general, the International Agency for Research on Cancer - a branch of WHO -
classifies the EMFs generated by radiofrequency devices (i.e., a wide set of equipment
including mobile BSs and mobile phones) as possible carcinogenic to humans, based
on an increase of brain cancer associated to the use of the wireless phones [?]. The
International Agency for Research on Cancer (IARC) also suggests that further
research needs to be performed [?], in order to thoroughly assess the long term of
the EMFs on the public health.
2.2 EMF Measurements in Cellular Networks
The measurement of EMFs generated by cellular networks allows to derive indi-

cations about the impact of BSs and personal devices, such as mobile phones, on
the total EMF absorbed by the body. Focusing on the EMFs generated by the BSs,
Koprivica et al. [?] perform a measurement campaign of 2G BSs, highlighting the
fact that the EMF tends to vary with the location, as well as the time of the mea-
surements. In addition, they also found different locations experiencing EMF levels
higher than the ones imposed by national limits. On the other hand, Urbinello et al.
[?] perform EMF measurements in different European outdoor urban environments,
concluding that the measured EMF are well above the national limits. Huang et al.
[?] perform a wide set of measurements in different areas located in Serbia and in
France, mainly focusing on 3G networks. Similarly to [?], they also observed a large
variability in the measurements across space and time. Fernández-García and Gil [?]
conduct a measurement campaign in an European city, observing EMF levels lower
than the maximum ones enforced by national law. Orłowski et al. [?] measure the
EMFs generated by different BSs located in Poland, concluding that the total EMF is
lower than the maximum limit in all cases except one. Summarizing, the considered
previous works study the EMFs generated by the BSs, showing that, in general, the
BS EMFs are lower than the maximum values imposed by law. This outcome is also
confirmed by the survey of Sagar et al. [?], which analyzes the research works about
EMF exposure in Europe during the period 2000-2015.
A second taxonomy of works regards instead the measurements of EMFs gen-
erated by the mobile phones. Joseph et al. [?] perform a comparison of the EMF
measured in proximity to users living in different European countries, finding that
the largest contribution to the EMF exposure is due to mobile phones used in trans-
port vehicles (e.g., trains, cars, buses). In particular, the EMF generated by mobile
phones is larger than the one received from the BS in all the environments, except
outdoor urban. Moreover, it is important to note that the study in [?] has been con-
ducted different years ago, when state-of-the-art 4G and 5G technologies were not
yet available. Frei et al. [?] examine the levels of exposure and the EMF sources for
a sample of volunteers living in a Swiss city, showing that the exposure to EMFs
notably varies across the different persons participating to the test. In addition, mo-
bile phones, BSs, and cordless phones are the main sources of exposure in urban
areas. More recently, Roser et al. [?] perform an assessment of the EMFs for a set of
adolescents living in Swiss. Interestingly, the EMFs generated by the mobile phones
dominates over all the other sources. Summarizing, mobile phones appear to be a
non-negligible contributor to the EMFs absorbed by users, with different evidences
suggesting that the EMFs generated by the mobile phones is larger than the one
received by the BSs or other sources.
2.3 Planning with EMF constraints
The planning of cellular network is a challenging problem [?], which needs to select
the locations of the sites which host BSs, as well as the configuration of each BS
in terms, e.g., of radiated power. The planning of cellular networks under EMF
constraints is even a more complex problem, as the presence of already deployed
BSs, as well as the EMF limits, severely impacts the obtained planning, in terms of
BS locations and their configuration [?]. In this context, several commercial tools (see
e.g. [?, ?, ?, ?, ?]) are used by operators and regulators in order to simulate the impact
of a given BS planning. These tools require a precise characterization of the scenario
in terms e.g, of digital elevation model, 3D buildings/obstacles, already deployed
networks, users/traffic distribution and EMF regulations. However, we stress the fact
that both the selection of the initial set of BS sites and their configuration have a
large impact on the obtained results [?].
3 EMF Limits in Italy and Their Effect on the Cellular Planning
EMF limits are thresholds on maximum EMF exposure, enforced by national regu-
lations to ensure that technologies emitting electromagnetic fields do not represent
a danger for the public wellbeing. In the specific case of 5G systems, radio access
networks will operate across a wide spectrum of frequencies, from hundreds of
MHz to tens of GHz, so as to support the variety of requirements entailed by the
many emerging mobile applications [?]. At 5G frequencies, EMFs are known to
induce mainly thermal effects (i.e., induced current, or skin and body heating) on
the human body; hence, the concerns above easily apply to next-generation cellular
deployments.
Many countries worldwide, as well as the European Council, adopt EMF limits set
by the ICNIRP [?]. Although ICNIRP limits are already fairly strict, Italy enforces
different EMF limits on its national territory [?], which are even more constraining.
Specifically, two distinct classes of limits are introduced by the italian law: (i)
general limits that are in most cases around 30% lower than the ICNIRP ones, and
5G Site Location 5G Site coverage
More stringent EMF limits

(a) Extensive 5G cell densification
Legacy
pre-5G
Site
Zone with EMF Saturation

No new 5G-site allowed
New
5G-Site
Service Area
(b) Presence of an EMF saturation zone
School
New
5G-Site
Minimum Distance
(c) EMF regulation based on a minimum distance
Fig. 1 Three situations where regulations on EMF limits impact the planning of 5G network
infrastructure [?]
(ii) restrictive limits that are 10 times lower than the ICNIRP ones. The restrictive
limits apply in fact to a vast portion of the national territory, including houses
(including terraces and balconies), schools, and in general to buildings for human
long-stay purposes. In addition, further regulations at local levels (see e.g., [?, ?])
establish minimum distances between BS sites and sensitive places (e.g., schools or
hospitals), as well as among base stations themselves. To cite an example, the city
of Rome has a very stringent regulation, which adopts the national limits, plus a
minimum distance of 100 [m] between a BS site and a sensitive place.
The current directives above jointly lead to a scenario where EMF limits risk
to have a significant impact on the deployment of 5G radio access infrastructures.
Indeed, EMF limits reduce the operators’ flexibility in installing new base station
sites in multiple ways. Three representative examples are outlined in Fig. 1. Fig. 1(a)
illustrates the cell densification phenomenon: stringent EMF limits impose low
radiation power at each antenna, which in turn forces a dense deployment of low-
power sites, with a substantial increase in CAPEX for the operator. Fig. 1(b) shows
a case where the 3G/4G sites serving the target region already saturate the EMF
limits, hence the operator is forced to install 5G antennas in a new site, with increased
CAPEX and reduced possibilities for network planning optimization. Fig. 1(c) recalls
situations where a minimum distance must be respected between a new 5G site and
a sensitive place, such as a school, again leading to sub-optimal network planning
and possibly curbed Quality of Experience (QoE) by the mobile service end users.
Those illustrated before are only a few, simple examples of the type of barriers
that EMF limits may pose to the planning and deployment of next-generation cel-
lular network radio access infrastructures. And, the presence of multiple competing
operators will only exacerbate the problem, which thus exposes 5G systems, as well
as the many and varied disruptive applications they are expected to enable, to serious
risks of under-performance.
4 Case Studies
We present two case studies that provide insight on two complementary aspects of
the impact of EMF limits on 5G network planning. In the first one we investigate the
problem of EMF level saturation, an issue that can severely hamper the transition
between pre-5G and 5G technologies. Then, we analyze the impact of restrictive
EMF regulations on network planning and user QoS.
4.1 Evidence of EMF saturation
In the first considered case study we consider a 1100×1100 [m2 ] zone in the Fuori-
grotta district, Naples. The San Paolo stadium and several buildings of the Engineer-
ing School of the University of Naples Federico II are located in the area, which
makes it a relevant test site, characterized by the presence of a huge number of users.
Five known BS sites are present in this area, operated by two of the main Italian
mobile operators, TIM and Wind Tre. Detailed information on the site configurations
are reported in [?]. This information, along with a description of the scene in terms of
(a) 100% maximum input power (b) 75% maximum input power
Fig. 2 Impact of the maximum input power (Fuorigrotta case-study) [?].
a Digital Elevation Model and of a vector file providing description of the buildings,
is used as input in a ray tracing simulator [?, ?]. Our purpose is to compare the over-
all EMF level produced by the antennas with Italian EMF limits: since the involved
frequencies (lower than 3 GHz) share the same Italian general limit of 20 [V/m], the
individual EMF intensities can be incoherently summed and a map of the overall
EMF amplitude level can be obtained taking the square root of this summation. Once
this map is available, it is possible to highlight those areas where EMF levels are
higher than the allowed limits. In Fig. 2.a these areas are marked in red: actually, in
most of the square located South of the stadium EMF levels are higher than 20 [V/m].
This result represents the worst-case scenario, obtained assuming that each antenna
keeps radiating the maximum declared power during the whole measurement time,
i.e. 6 minutes and 24 hours according to ICNIRP and Italian guidelines, respectively.
This kind of approach is based on the fact that in principle no limit is imposed on
mobile operators, which are authorized to constantly radiate the maximum declared
power: note that this worst-case scenario approach is frequently assumed for EMF
level evaluation [?, ?, ?]. Moreover, note that the measurement procedures used to
verify the compliance of EMF levels not only change for different countries, but in
a single country are likely to be revised, especially when 5G systems will begin to
operate. In this situation, the proposed approach is useful to study EMF saturation
regardless of the specific national laws. However, since the assumption that all the
considered antennas constantly radiate the maximum power may be too strong, in
Fig. 2.b we report the area where EMF limits are violated assuming that each antenna
is radiating 75% of the maximum declared power: even in this case EMF limits are
are exceeded in many parts of the scene.
The presented analysis confirms that EMF saturation may be a relevant issue for
the deployment of future 5G sites, since in some scenario EMF values very close to
the prescribed limits may have been reached by pre-5G networks. However, some of
the new technologies associated to 5G networks, such as beamforming, may be used
RSRP [dBm] RSRP [dBm]
106 106
-50 -50
4.6315 4.6315
-60 -60
-70 -70
4.631 -80 4.631 -80
-90 -90
[m]
[m]
4.6305 -100 4.6305 -100
-110 -110
4.63 -120 4.63 -120
-130 -130
4.6295 -140 4.6295 -140

2.855 2.86 2.865 2.87 2.875 2.855 2.86 2.865 2.87 2.875
[m] 105 [m] 105
(a) TIM (b) Vodafone

106
RSRP [dBm]
-50
4.6315
-60
-70
4.631 -80
-90
[m]
4.6305 -100
-110
4.63 -120
-130
4.6295 -140
2.855 2.86 2.865 2.87 2.875
[m] 105
(c) Wind Tre
Fig. 3 RSRP metric (TMC case-study) [?].
to implement a smart planning of EMF levels over the area of interest, providing a
potential way to limit the effects of EMF saturation. Some hints on this topic will be
provided in Section 5.
4.2 Impact of EMF limits on planning and QoS
In this case study we focus on the area "Torrino-Mezzocammino" (TMC), Rome,

Italy. It is a residential neighborhood, including commercial buildings, schools, pub-
lic parks, with a population of more than 10000 people. In this area a local regulation
dictates a minimum distance (100 [m]) between BS and "sensitive places", whose
definition is referred to the municipality in accordance with the local population
(e.g., highly frequented sport centers and commercial buildings have been marked
as "sensitive"). In this situation, our aim is to investigate how local regulations im-
pact on network planning and user QoS. In this analysis we used the CellMapper
application [?] to measure relevant cellular network parameters, with associated ge-
ographic position data, while traveling along TMC area. Several interesting results
were obtained from our analysis. First, thanks to CellMapper we identified the type
of service offered by mobile operators in the TMC area: most of the area is served
by a 4G LTE service, with some areas covered by LTE-Advanced and some other
with a 3G HSPA+ service. As a further step, we provide an estimate of the QoS
based on the Reference Signal Received Power (RSRP) metric. In particular, thanks
to CellMapper we are able to measure the RSRP of the current BS in the TMC area.
Averaging the obtained values in 30x30 [m2 ] cells allowed us to obtain a 30 [m] res-
olution RSRP map of the area of interest. This is done for each operator. Since there
is a relantionship among the RSRP and the Signal to Noise Ratio (SNR) [?], this map
can be considered as a good indicator of the quality of the radio link between the
User Equipment (UE) and the BS. The RSRP maps obtained for the three operators
are shown in Fig. 3, which testifies that some zones present values of RSRP lower
than −110 [dBm], associated to a very low user QoS. This is confirmed by frequent
drops of the radio link and difficulties in the access to internet services. Indeed, many
of the areas experiencing good values of the RSRP are located in zones served by
BSs installed outside the TMC area.
This case study demonstrates how an insufficient number of BS sites in the TMC
area impacts both on the type of service provided by the operators and on the user
QoS. Moreover, it shows how the coexistence of many heterogeneous regulations,
not only at the international and national scales, but even at the local scale, severely
hampers the deployment of 5G networks, limiting the possibility of developing
general solutions to the planning problem.
5 Expected Impact of 5G Features on EMF
As indicated in [?], 5G will be characterized by a set of radically new technologies.

In this Section, we analyze the main ones in term of relevance and expected impact
on the EMF exposure levels compared to current deployed cellular networks. Clearly,
the developed analysis is not exhaustive and it needs to be supported and verified via
extensive EMF measurement campaigns.
Tab. 1 summarizes each 5G technology feature, as well as its relevance in terms
of EMFs. To this aim, the utilization of massive MIMO antenna arrays is one of
the recognized 5G features. In turn, this implies an increased number of antennas
radiating power at each site. Although some recent works [?, ?] suggest that this 5G
feature will impact positively on the EMF exposure levels, the reasonings in [?, ?]
cannot be generalized. In fact, the MIMO impact highly depends on the selected
configurations and on the adopted approach for measuring the EMF levels [?, ?]. As
a consequence, the academic and industrial communities need to assess in the near
future the potential MIMO gain in terms of EMF levels.
Beamforming constitutes another recognized 5G feature. It allows spatial se-
lective communications, since the radiated power is concentrated into narrow and
selected spatial directions. As a consequence, it is believed that, compared to current
deployed BSs, beamforming will induce a general decrease of the EMF exposure
levels. However, due to the power concentration into selected directions, there may
be an increase of the EMF levels in the portions of the territory corresponding
to these directions. Similarly, the exploitation in 5G networks of mmWave bands
for directional communications may be beneficial for the EMF exposure levels. In
fact, mmWave communications are characterized by higher path-losses compared to

the current micro-waves communications. Hence, the received EMF may be lower
compared to the one generated by currently deployed BSs. In this context, it is
worthwhile to mention that 5G is envisioned to exploit different coexisting cell tiers
with a dense deployment of small cells. Hence, it is expected a decrease of the EMF
exposure levels compared to the current deployed BSs, due to the shorter BS-to-user
distances. However, a possible increase of the EMF levels in proximity to the small
cell may be also experienced when comparing to the current deployment. The EMF
level decrease, promised by the small cell exploitation, is expected to be further
enhanced by adopting offloading mechanisms, enabled by the unavoidable coverage
overlapping among the different cell tiers. In fact, according to such mechanisms,
the power radiated from the most loaded cells could be reduced by offloading the
users to other cells in proximity to them.
Other key 5G technologies envisioned to decrease the EMF exposure levels are
the softwarization and the Mobile Edge Computing (MEC). Specifically, on one
hand with the softwarization, since different network functionalities, including most
of the BS functionalities, will be realized at software level, it will be possible
that multiple operators will share the same BS hardware. As a consequence, the
number of antennas radiating power and installed in the same site will decrease with
respect to the case in which each operator installs its own physical equipment in
the shared site. On the other hand, with the MEC, cloud computing capabilities will
be enabled at the edge of the network. By properly managing the content stored in
the MEC platforms, it may be possible to decrease the amount of transferred data,
thus decreasing the radiated power. However, this vision will be affected also on
the type of service provided by MEC, which may include high data rate services as
augmented reality. Similarly, the adoption of the Device-to-Device (D2D) paradigm
in 5G networks is envisioned as a way to decrease the EMF exposure levels. In fact,
with D2D a decrease in the amount of information exchanged between the UE and
the BS will be expected. This in turn will reduce the amount of EMFs generated
by the BS. In the same direction, advanced power saving techniques, including
deep Sleep Modes (SMs), may be exploited in 5G networks. Specifically, SM-based
techniques may reduce the amount of EMF exposure levels, since the BS that are
not used are completely switched off (or in a low-power state). However, there could
be an increase in the EMF levels in proximity to the BSs that remain powered on
and have to increase their coverage also to the zones previously served by the BSs
currently in SM. Finally, with the deployment of 5G networks, legacy 2G and 3G
networks will be dismissed. Clearly, this will positively impact on the EMF exposure
levels, especially in terms of reduction of the current EMF saturation levels in urban
environments.
Clearly, the advantages and the disadvantages in terms of EMF exposure levels
forecasted for the described 5G technologies need to be assessed against the aggre-
gate radiation generated by the different BS layers operating with different access
technologies, especially if legacy pre-5G networks already showed levels of EMF
saturation, as revealed by our case-studies. Moreover, they need to be assessed against
the dependence of the EMF exposure levels on several factors, including e.g., the
type of BS/UE, the BS/UE location with respect to the user, and the location of the
user [?].
Table 1 Expected Impact of the 5G Technology Features on the EMF levels [?].
Feature Relevance to EMFs Expected EMF Increase/Decrease
MIMO Increased number of antennas radiat- -/+ The impact on the EMFs levels de-
ing power. Impact of computing the ra- pends on the specific MIMO configu-
diated power when assessing the com- ration and on the adopted approach for
pliance with EMF limits. measuring the EMF levels.
Beamforming Directionality control of the radiated - General decrease w.r.t. currently de-
power. ployed BSs.
Power concentrated into selected loca- + Increase in selected locations.
tions.
mmWave Path loss increase of radiated signals - (Possible) decrease w.r.t. BSs ex-
on mmWave bands. ploiting micro-waves.
Small Cells Installation of additional sources of - (Possible) decrease w.r.t. the current
power. Less power required to macro cellular network.
cells.
+ (Possible) increase in proximity to
the small cells.
Offloading (Possible) reduction of radiated power - (Possible) decrease w.r.t. the current
from the most loaded cells. cellular network.
Softwarization Sharing of the hardware infrastructure - Large decrease w.r.t. the case in
by multiple operators. Less antennas which each operator installs its own
installed in the shared sites. physical equipment in the same site.
MEC (Possible) decrease in the amount of - (Possible) decrease w.r.t. to the cur-
transferred data in the air, thus de- rent MEC-unaware network.
creasing the radiated power.
D2D Reduction of tha amount of data trans- - Decrease w.r.t. current deployments
ferred (and consequently of power) be- exploiting classical communication
tween the BS and the UE. schemes (e.g., UE to BS).
Sleep mode BSs put in sleep mode radiate zero (or - Decrease in proximity to the BSs put
very low) power. in sleep mode.
The BSs that remain powered on may + (Possible) increase in proximity to
have to increase their coverage area. the BSs that remain powered on.
2G/3G Dismission Reduction of the current EMF satura- - Large decrease w.r.t the case in which
tion levels in urban zones. all the legacy technologies are main-
tained.
6 Conclusions and an outlook
In this work we focused on the planning of 5G networks under EMF exposure con-
straints. A detailed analysis of the state of the art regarding the EMF-related health
risks, EMF measurement campaigns, and the impact of EMF limits on cellular net-
work planning was presented. The effects of Italian EMF limits on cellular planning
were considered and our observations were supported by two real-world Italian case
studies. More in detail, these case studies provided evidences that: (i) the installation
of 5G sites may be hampered by EMF saturation effects already observed in pre-5G
sites; (ii) the sub-optimal planning due to the restrictive regulations on EMF expo-
sure has a negative impact on the type of provided service as well as on the QoS
perceived by the user. Finally, we analyzed the potential impact of the main 5G tech-
nologies on the EMF exposure levels. The presented analysis is not exhaustive and
it needs to be verified and supported via extensive measurement campaigns. From
the developed discussion, the complex nature of 5G planning under EMF constraints
clearly emerges. However, this is a challenge that cannot be missed, if we want to
avoid negative effects on the side of both the mobile operators (in terms of CAPEX
increase) and the final users (in terms of perceived QoS).
The role of propagation software tools for
planning 5G wireless networks
Antonio Iodice, Daniele Riccio, Giuseppe Ruello 1
Abstract The demand for faster and faster data transfer, and the development
of the Internet of things (IoT) and of machine-to-machine (M2M)
communications, are urging toward a new generation of wireless networks,
namely, the fifth generation (5G). The most obvious way to obtain the
necessary bandwidth is to move towards higher frequencies, so that use of
millimeter (mm) waves is being considered. Accordingly, a strong effort is
being made by researchers to characterize propagation of mm waves in
complex environments, such as urban areas. Here we show how use of a
software tool for the prediction of electromagnetic propagation in complex
scenarios can help the design of 5G wireless networks. As a first example,
electromagnetic propagation in urban areas at 3.7 and 28 GHz, two
frequencies currently employed for the first preliminary experiments of 5G
networks in Italy and in USA, is here analyzed by using the software tool.
Antonio Iodice
Università degli Studi di Napoli Federico II
Daniele Riccio
Giuseppe Ruello
246 A. Iodice, D. Riccio, G. Ruello
1 Introduction
The design of the fifth generation (5G) mobile network is bringing
growing concern on the need of accurate, efficient, flexible and scalable tools
for the prediction of the electromagnetic field propagation [1]–[3].
In fact, the 5G paradigm must fulfill the huge demand of bandwidth
needed to provide real time, multimedia services and this goal will be reached
by exploiting new design concepts. Key issues will be the reduction of the cell
size, the use of mm waves and the use of MIMO technology [1], [4]–[7]. Each
of these issues requires a deep comprehension of the physical phenomena that
govern the propagation of the electromagnetic field, mainly in complex
environments, as urban areas.
The use of the millimeter-wave spectrum poses new challenges to the
network designers, who will face limited propagation ranges, increased
atmosphere attenuation and high sensitivity to events that can cause service
blocks. In this context, the use of a radiopropagation software with favorable
trade-off between computational time and accuracy is crucial.
In the open literature, several electromagnetic propagation solvers can be
found [8]–[24]. Proposed strategies and frameworks differ mostly for the
selected models of the propagation channel. Urban areas are described by
appropriately superimposing simple (fixed or moving) elemental objects
(buildings, roads, cars, etc…). These objects can be modeled partly by means
of deterministic parameters (usually, geometric coordinates, average building
shape, mean roughness and dielectric constant) and partly by means of
stochastic contributions (appropriate for building details, moving objects,
surfaces roughness etc…). Accordingly, the field propagation is modeled by
means of deterministic and random as well as time-invariant and time-varying
contributions. Summarizing, the solvers are usually classified as probabilistic
[17]–[22], if they consider the city environment as a representation of a
stochastic process, and deterministic [8]–[16] if they provide a specific site
propagation model.
Probabilistic models provide general analytic results on the average
properties of the field propagation. They are suitable for predicting on average
the radio coverage, but they lack specificity and accuracy.
If a specific site is of interest, high frequency descriptions of the
electromagnetic field can be safely used to describe microcellular propagation
in a deterministic environment: ray-tracing and ray-launching techniques are
commonly employed. Despite of their high computational complexity, these
methods are evolving toward significant performances in terms of precision
and computational time [8]–[16]. Most of them reduce the computational time
Role of propagation tools 247
by a priori limiting the number of interactions between the rays and the scene
(modeled as reflections and diffractions).
In this Chapter we illustrate the use of an electromagnetic solver,
previously developed by the authors [22] for the evaluation of the
electromagnetic coverage in urban environment, as a tool for planning 5G
wireless networks. The solver implements a ray-launching technique,
following the footsteps of the pioneer work by Liang and Bertoni [8].
The tool is precise from the electromagnetic viewpoint because any
significant electromagnetic contribution is taken into account. Line of sight
links, reflections, diffractions, ground scattering and transmission are
accounted for in closed form. Innovative fractal models [30] guarantee the
efficient and precise description of diffuse scattering from natural surfaces.
By means of simple information on the ray optical path, the employed ray-
launching technique allows also evaluating over the considered area the pulse
delay and spreading information. The tool is written in IDL (Interactive Data
Language). With this choice, computational efficiency is slightly sacrificed in
favor of portability on different hardware platforms. The flexibility of the tool
allows its use in several scenarios, including not only mobile network
planning, but also, as instance, wireless sensors networks [25] and Internet of
Things.
As a first example of use of the tool, electromagnetic propagation in urban
areas at 3.7 and 28 GHz, two frequencies currently employed for the first
preliminary experiments of 5G networks in Italy and in USA, is here analyzed.
In fact, his year, some mobile phone companies in Italy are starting an
experimentation of 5G networks to test the achievable quality of service (QoS)
[26]. This experimentation is performed at 3.7 GHz (the frequency band
released by the Italian Government for this purpose), but a possible aim is to
try to “scale” obtained results also to the higher, 28 GHz, band, for which
some results are available in recent literature [27].
2 Electromagnetic solver
The electromagnetic solver input is a digital description of the scene, see
Fig. 1a, and of the transmitting antenna, see Fig. 1b. The scene description is
provided by a vector file in Planet or kml (Keyhole Markup Language)
formats describing the buildings, and a raster file describing the terrain
topography (Digital Terrain Model, DTM). Buildings' walls and terrain
relative permittivity and conductivity can be also stored to account for the
electromagnetic properties, that are frequency dependent.
A vertical-plane-lunching (VPL) raytracing algorithm [8],[22] is
employed that considers direct, reflected and diffracted rays. Reflections are
treated by using Geometrical Optics (GO) [28], whereas diffraction is
evaluated by using the Uniform Theory of Diffraction (UTD) [29]. The
software tool solves the trade-off between speed and accuracy, without
determining a priori the number of considered reflections and diffractions.
The tool stops propagating each ray when its amplitude goes below a threshold
that can be independently set according to the user needs. Such a criterion
provides speed and accuracy performances adequate for cell planning.
The electromagnetic field is computed on one or more regular 2-D grids
(“layers”) placed on surfaces at different fixed heights above the ground (or
above the rooftop, if the grid point is in correspondence of a building), see
Fig.2a.
The employed raytracing solver is also able to compute delay spread and
angular spread, that are visually displayed by using a “radar-like”
representation, see Fig. 2b.
(a) (b)
Fig. 1. Inputs of the software tool: (a) 3D representation of the input scenario, where the
red spot identifies the antenna position, and (b) antenna description, with horizontal and vertical
radiation diagram cuts.
(a) (b)
Fig. 2. Outputs of the tool: (a) Color scale representation of the intensity of the field predicted
by the software, and (b) “radar-like” representation of the contributions of all the rays to the
field in a specific location; the color and the distance of the arrow from the center indicate the
field intensity and the delay, respectively.
C C
B B
A A
(a) (b)
Fig 3. Received signal strength L, in dBm, in the area of Piazza Plebiscito, Naples, Italy. The
(omnidirectional) transmitting antenna is represented by a red diamond. Transmitted power is
5 W. (a) 3.7 GHz (black: L<-100 dBm, red: L>-40 dBm). (b) 28 GHz (black: L<-100 dBm,
red: L>-60 dBm).
3. Results
As anticipated in the Introduction, as a first example of use of the tool,
electromagnetic propagation in urban areas at 3.7 and 28 GHz is here
analyzed. Differences between the propagation results at the two considered
frequencies may be due, of course, to the different values of electromagnetic
parameters of building walls and soil. Another significant difference is the
increase of scattered and absorbed power by small obstacles (cars, people,
etc.) and by vegetation at the higher frequency. However, the main
dependence in the line-of-sight (LoS) case is expected to be the free-space
power loss increase with the square of frequency (for fixed-gain receiving
antennas). In the non-line-of-sight (NLoS) case, where the main propagation
mechanism is diffraction, a further power loss increase proportional to
frequency is added, so that an overall loss increase with the cube of frequency
is expected. This theoretical expectation is mostly confirmed by solver
simulations. For instance, in Fig. 3 simulated field levels in an area in the
Naples city center at 3.7 (a) and 28 (b) GHz are shown. In the area A,
characterized by a LoS link, an average difference of about 18 dB
(corresponding to the square of the frequency ratio) between results at the two
frequencies is obtained; and in the NLoS area B, in which the dominant
mechanism is diffraction, an average difference of about 27 dB
(corresponding to the cube of the frequency ratio) is present. Finally, in the
NLoS area C, characterized by the so-called “canyoning” effect, in which both
reflections and diffractions play a role, an intermediate average difference of
about 22 dB is obtained.
It must be finally noted that, while atmospheric gases do not appreciably
affect propagation at any of the two considered frequencies, attenuation by
rain may significantly affect propagation at 28 GHz, see Fig. 4, where
attenuation as a function of rain rate at 3.7 and 28 GHz is plotted by using the
empirical formulas of Olsen, Rodgers and Hodge, as reported in [28], page
404. The optional possibility to account for this further attenuation has been
added to the solver of [22].
Fig 4.
Attenuation by
rain (in dB/km)
as a function of
rain rate (in
mm/h) at 3.7 and
28 GHz.
4. Conclusion
We have described an electromagnetic solver able to predict propagation
in a complex scenario, such as an urban area, and we have illustrated its use
to help design of 5G wireless networks. In particular, electromagnetic
propagation in urban areas at two frequency bands of interest for 5G wireless
networks has been considered. Main sources of differences between
propagation behaviors in urban areas at the two considered frequencies have
been briefly analyzed.
Performed analysis has only considered the received signal strength.
However, we have shown how the employed raytracing solver is also able to
compute delay spread and angular spread: the former also affect the final QoS,
and the latter can be exploited to devise the use of reconfigurable directive
receiving antennas.
References
1. J. G. Andrews, S. Buzzi, W. Choi, S. V. Hanly, A. Lozano, A. C. K. Soong, and J. C.
Zhang, “What will 5G be?,” IEEE J. Sel. Areas Commun., vol. 32, no. 6, pp. 1065–1082,
2014.
2. T. S. Rappaport, S. Sun, R. Mayzus, H. Zhao, Y. Azar, K. Wang, G. N. Wong, J. K.
Schulz, M. Samimi, and F. Gutierrez, “Millimeter wave mobile communications for 5G
cellular: It will work!,” IEEE Access, vol. 1, pp. 335–349, 2013.
3. M. Agiwal, A. Roy, and N. Saxena, “Next Generation 5G Wireless Networks: A
Comprehensive Survey,” IEEE Commun. Surv. Tutorials, vol. 18, no. 3, pp. 1617–1655,
2016.
4. P. Kántor, J. Bitó, and Á. Drozdy, “Characteristics of 5G wireless millimeter wave
propagation: Transformation of rain attenuation applying different prediction models,”
2016 10th Eur. Conf. Antennas Propagation, EuCAP 2016, 2016.
5. A. I. Sulyman, A. Alwarafy, G. R. MacCartney, T. S. Rappaport, and A. Alsanie,
“Directional Radio Propagation Path Loss Models for Millimeter-Wave Wireless
Networks in the 28-, 60-, and 73-GHz Bands,” IEEE Trans. Wirel. Commun., vol. 15, no.
10, pp. 6939–6947, 2016.
6. M. R. Akdeniz, Y. Liu, M. K. Samimi, S. Sun, S. Rangan, T. S. Rappaport, and E. Erkip,
“Millimeter wave channel modeling and cellular capacity evaluation,” IEEE J. Sel. Areas
Commun., vol. 32, no. 6, pp. 1164–1179, 2014.
7. S. Rangan, T. S. Rappaport, and E. Erkip, “Millimeter-wave cellular wireless networks:
Potentials and challenges,” Proc. IEEE, vol. 102, no. 3, pp. 366–385, 2014.
8. G. Liang and H. L. Bertoni, “A new approach to 3-D ray tracing for propagation
prediction in cities,” IEEE Trans. Antennas Propag., vol. 46, no. 6, pp. 853–863, 1998.
9. J. Walfisch and H. L. Bertoni, “A Theoretical Model of UHF Propagation in Urban
Environments,” IEEE Trans. Antennas Propag., vol. 36, no. 12, pp. 1788–1796, 1988.
10. V. Degli-Esposti, D. Guiducci, A. de’Marsi, P. Azzi, and F. Fuschini, “An advanced
field prediction model including diffuse scattering,” IEEE Trans. Antennas Propag., vol.
52, no. 7, pp. 1717–1728, 2004.
11. F. Fuschini, H. El-Sallabi, V. Degli-Esposti, L. Vuokko, D. Guiducci, and P.
Vainikainen, “Analysis of multipath propagation in urban environment through
multidimensional measurements and advanced ray tracing simulation,” IEEE Trans.
Antennas Propag., vol. 56, no. 3, pp. 848–857, 2008.
12. X. Gao, J. Zhang, G. Liu, D. Xu, P. Zhang, Y. Lu, and W. Dong, “Large-scale
characteristics of 5.25 GHz based on wideband MIMO channel measurements,” IEEE
Antennas Wirel. Propag. Lett., vol. 6, pp. 263–266, 2007.
13. M. C. Lawton and J. P. McGeehan, “The Application of a Deterministic Ray Launching
Algorithm for the Prediction of Radio Channel Characteristics in Small-Cell
Environments,” IEEE Trans. Veh. Technol., vol. 43, no. 4, pp. 955–969, 1994.
14. W. M. O’Brien, E. M. Kenny, and P. J. Cullen, “An efficient implementation of a three-
dimensional microcell propagation tool for indoor and outdoor rban environments,”
IEEE Trans. Veh. Technol., vol. 49, no. 2, pp. 622–630, 2000.
15. H. H. Xia, H. L. Bertoni, L. R. Maciel, A. Lindsay-stewart, and R. Rowe, “Microcellular
Propagation Characteristics for Personal Communications in Urban and Suburban
bnvironment s,” vol. 43, no. 3, 1994.
16. H. R. Anderson, M. Sc, and P. Eng, “A Ray-Tracing Propagation Model for Digital
Broadcast Systems in Urban Areas,” vol. 39, no. 3, pp. 309–317, 1993.
17. G. Franceschetti, S. Marano, and F. Palmieri, “Propagation Without Wave Equation
Toward an Urban Area Model,” IEEE Trans. Antennas Propag., vol. 47, no. 9, pp.
1393–1404, 1999.
18. S. Marano and M. Franceschetti, “Ray propagation in a random lattice: A maximum
entropy, anomalous diffusion process,” IEEE Trans. Antennas Propag., vol. 53, no. 6,
pp. 1888–1896, 2005.
19. A. Bender and T. Rainsford, “Inhomogeneous Markov approach to percolation theory
based propagation in random media,” IEEE Trans. Antennas Propag., vol. 56, no. 10, pp.
3271–3284, 2008.
20. D. Chizhik and J. Ling, “Propagation over clutter: Physical stochastic model,” IEEE
Trans. Antennas Propag., vol. 56, no. 4, pp. 1071–1077, 2008.
21. N. Blaunstein, M. Toeltsch, C. G. Christodoulou, J. Laurila, E. Tsalolihin, E. Bonek, P.
Vainikainen, N. Tsouri, K. Kalliola, and H. Laitinen, “Azimuth, elevation, and time-
delay distributions in wireless communication channels,” IEEE Antennas Propag. Mag.,
vol. 48, no. 1, pp. 160–167, 2006.
22. G. Franceschetti, A. Iodice, D. Riccio, and G. Ruello, “A Tool for Planning
Electromagnetic Field Levels in Urban Areas,” in Proc. of the IEEE Antennas and
Propagation Society International Symposium, 2004, pp. 2211–2214.
23. S. Hussain and C. Brennan, “An efficient ray tracing method for propagation prediction
along a mobile route in urban environments,” Radio Sci., vol. 52, no. 7, pp. 862–873,
2017.
24. J. H. Lee, J. S. Choi, and S. C. Kim, “Cell Coverage Analysis of 28 GHz Millimeter
Wave in Urban Microcell Environment Using 3-D Ray Tracing,” IEEE Trans. Antennas
Propag., vol. 66, no. 3, pp. 1479–1487, 2018.
25. C. Antonopoulos, K. Asimogloy, S. Chiti, L. D’Onofrio, S. Gianfranceschi, D. He, A.

Iodice, S. Koubias, C. Koulamas, L. Lavagno, M. Lazarescu, G. Mujica, G.
Papadopoulos, J. Portilla, L. Redondo, D. Riccio, T. Riesgo, D. Rodriguez, G. Ruello, V.
Samoladas, T. Stoyanova, G. Touliatos, A. Valvo, and G. Vlahoy, “Integrated Toolset
for WSN Application Planning, Development, Commissioning and Maintenance: The
WSN-DPCM ARTEMIS-JU Project,” Sensors, vol. 16, no. 6, p. 804, 2016.
26. http://www.sviluppoeconomico.gov.it/index.php/it/198-notizie-stampa/2037116-5g-in-5-
citta-al-via-la-sperimentazione-della-tecnologia-mobile-di-v-generazione , accessed on
April 3, 2018.
27. T. S. Rappaport, G. R. MacCartney, M. K. Samimi, S. Sun, “Wideband Millimeter-Wave
Propagation Measurements and Channel Models for Future Wireless Communication
System Design,” IEEE Transactions on Communications, vol. 63(9), pp. 3029-3056,
Sep. 2015.
28. R. E. Collin, Antennas and Radiowave Propagation, McGraw-Hill, New York, 1985.
29. R. G. Kouyoumjian and P. H. Pathak, “A uniform GTD for an edge in a perfectly
conducting surface,” Proc. IEEE, vol. 62, no. 11, pp. 1448–1461, 1974.
30. G. Franceschetti and Riccio, Scattering, Natural Surfaces and Fractals, Academic P.
Burlington, MA, 2007.
Photonics for 5G
Antonella Bogoni, Luca Potì, Giancarlo Prati, Marco Romagnoli1
Abstract Photonic technologies are largely used in optical communication

systems and networks due to their unique characteristics in term of bandwidth,
immunity to electromagnetic fields, compatibility with the optical fiber and
flexibility. The possibility to fabricate photonic devices through CMOS
compatible processes paves the way for low cost and reduced footprint circuits
making optical technologies available for any network segments and new
applications. Here the importance of Photonics for 5G development and
deployment is addressed together with some technological alternatives and
perspectives.
1 Optical systems enabling 5G
Antonella Bogoni
Scuola Superiore Sant’Anna - Pisa
Luca Potì
Photonic Networks & Tecnhologies Nat’l Lab – CNIT, Pisa
Giancarlo Prati
IRCPhoNeT, Pisa
Marco Romagnoli
256 A. Bogoni, L.Potì, G. Prati, M. Romagnoli
The rising penetration of smart connected devices, promising compelling services

anywhere and anytime, is having a huge impact on the mobile broadband
infrastructure and on the ability to provide a quality experience to the final users,
whether humans or connected things. The new 5G radio access ecosystem,
expected by 2020 [1], will have to sustain an average traffic increase of one order
of magnitude and peak rates up to three orders of magnitude higher than the
current ones [2]. Many time sensitive services will also demand extremely low
transmission latency. Radio technologies will evolve by allocating new bands
(beyond 10 GHz), by leveraging higher-order multiple-input multiple-output
(MIMO), carrier aggregation, and beam forming techniques. In parallel, the
transport network for backhaul (BH) and fronthaul (FH) applications will demand
support of higher capacities to increase the number of transport clients, to support
a wider range of performance requirements, and to provide increased flexibility.
This must be achieved in a cost-effective and sustainable manner. Based on the
radio architectures, it is possible to define many deployment scenarios, ranging
from the fully centralized one, i.e., the cloud radio access network (CRAN), to the
conventional scenario, in which all functions are replicated at each radio site with
monolithic radio base stations (RBSs). Moreover, as envisioned in [3], novel
radio-splitting models are under development to meet 5G high-bandwidth
demands, leveraging different distributions of radio functions between radio unit
nodes and centralized processing nodes. Realistic scenarios will see a mix of all
mentioned radio architectures, with a combination of traffic types to be
transported among the radio devices.
If mobile as communication was conceived for humans (mobile broadband and
media delivery), 5G will include also machine type communication between
devices (sensors, actuators etc.) that must communicate with the same wireless
access technology. A vast variety of use cases are predicted for 2020 each with
very different connectivity requirements: from the broadband access in a dense
area or in a crowd, monitoring and automation of infrastructures and buildings,
remote controlled heavy machines, control in real time of remote machines,
intelligent transportation system, communication with high user mobility (500
Km/h), etc. as shown in Fig. 1.
Fig 1.
5G use cases
Photonics for 5G 257
Fig 2.
Xhaul concept
The conventional point-to-point fronthaul concept is evolving towards a

geographical network connecting a pool of DUs with a plurality of RRUs using
the CPRI protocol (Xhaul concept shown in Fig. 2). Centralization of radio
baseband processing functions is gaining great interest for its potential to allow a
consolidation of nodes and network elements, so as to lower CapEx and OpEx
(e.g. fewer nodes to install, to maintain, to upgrade, and to power supply), while
at the same time increase radio coordination functions. Optical technologies with
their conventional benefits of high bandwidth, protocol transparency, scalability,
low latency, high resilience and network re-configurability, are today perceived
as a promising key piece of the radio access network puzzle, in both fronthaul and
backhaul transport areas. But previous generation of optical networking
technologies (e.g. SDH/SONET, WDM, OTN etc.), based on discrete components
and modules, that played a relevant role to realize an affordable transport medium
in metro and core networks are not adequate for the needs of the emerging RAN
transport segments requiring low cost, lower power consumption and a level of
miniaturization. Re-configurability features, provided by WDM technologies, can
further increase CPRI transport efficiency.
However, to use WDM approach in radio access networks it is fundamental to
dramatically reduce the cost figure of photonic components, modules, and sub-
systems, with respect to corresponding ones that are today used in metro
aggregation and long haul networks, to be acceptable in this portion of the
network. Such a cost reduction must be achieved while re-shaping the
performance of those ones. As an example, in a radio access network the
transmission span ranges from few hundreds of meters up to few tens of Km,
certainly not the several hundreds of Km that are required in metro and regional
networks. On the other hand, in mobile fronthaul segments it is a must to comply
with tight requirements in terms of latency, which could range in order of
magnitude of hundreds or even tens of milliseconds. As a result, to adopt WDM
technology in radio access and benefit of its peculiarities, it is necessary to reduce
cost and re-shape performance of photonic technology. Photonic Integration and,
in particular, silicon photonics with its recent advances in integrating many optical
circuits and functions (for instance multiplexer, attenuator, switches, couplers) in
a single chip using the well-developed CMOS production infrastructure, is the
ideal technology to fit the RAN needs.
In addition, when an RF signal is loaded on an optical carrier, it becomes
convenient to use optical fibers to transport the signal, implementing a Radio-
over-Fiber (RoF) system. In fact, fiber transmission is broadband, low-loss (as
low as 0.2 dB/km, while RF waveguides have a propagation loss in the order of
several dB/m), and EMI free. Therefore, the optical fiber allows transporting the
RF signal over long distances without significant distortions.
Considering the RoF system applied to a 5G radio system, the electro-optic and
opto-electronic signal conversions can be realized by straightforwardly
modulating the RF signal on an optical carrier at the optical transmitter side, and
then detecting the optical signal in a wideband photodiode at the optical receiver
side. If the fiber link is intended to remote a receiving antenna, as in the case of a
radar system, it can be convenient to keep the laser at a base station, send the
unmodulated optical carrier to the antenna through an optical fiber, and move to
the antenna only an optical modulator, where the carrier is loaded with the
received RF signal (Fig. 3) [4].
Fig 3.
Scheme of a RoF for remoting a receiver antenna.
Legenda Base station Antenna site
Optical
Electrical
Laser
MZM
RF signal from
the antenna
RF signal PD
A couple of possible issues arises with RoF solutions. The first one is related to
the available linear dynamic range of the transport system, in fact the RoF systems
using external optical amplitude modulation are often subject to nonlinearities
induced by the modulator itself. In order to increase the maximum linear dynamic
range, RoF systems exploiting phase modulation (PM) has been developed. While
this kind of modulation is very linear, the opto-electronic conversion of the RF
signal becomes more complex, requiring either a frequency discriminator for
implementing a direct detection, or an even more complex coherent detection
scheme [4]. A second issue is the effect of chromatic dispersion when the RF
signal is transferred on the optical carrier by means of a double sideband (DSB)
amplitude modulation. In fact, if the total chromatic dispersion of the fiber link is
high enough (for example, due to a long fiber link), some frequency components
of the two sidebands can undergo a significantly different phase shift due to the
dispersion, which gives a notch in the transmissivity of the system at those
frequencies that are turned exactly out-of-phase. In order to suppress the fading,
it is possible either to make use of phase modulations, or to exploit the single
sideband (SSB) amplitude modulation. Once the effect of the fading is suppressed,
RoF systems can be used to cover spans as long as several hundred km, also thanks
to the use of optical amplifiers (EDFAs).
Another application where optical technologies play a significant role is
represented by Optical beam forming. The beam forming of RF signals in phased
array antennas (PAAs, also called active electronically-steered antennas, AESAs)
allows steering the transmitted RF beam without physically moving the antenna.
This solution is used in wireless communications since it permits a strong
reduction of size and weight in the antennas (no moving parts are needed!). It is a
key functionality in 5G since next generation of mobile communication networks
will guarantee high throughput to the users, relying on a better efficiency in
resources allocation and power consumption, with a more extended use of the
frequency spectrum. User-specific beam forming (BF) will play a key role in the
actualization of this strategy, especially in very crowded environments, where a
precise steering and a fast re-orientation of the antenna beams are required [5].
In the next-future mobile and wireless communications scenario, beam forming
networks (BFNs) are expected to drive PAAs with potentially several elements,
whose size will shrink due to the employed high frequency. Therefore, antenna
pointing accuracy, low losses, reduced power consumption, and small size are
crucial characteristics in future BFNs. In PAAs, the time of emission of the signal
at each antenna element is controlled so that the wavefront of the signal generated
by the entire antenna array is synthesized to propagate in the desired direction.
The time of emission of the signal at each antenna element is commonly
controlled by means of electronic phase shifters (PSs), since a phase shifting is a
correct approximation of a time delay if the delay is small and the signal
bandwidth is much narrower than the carrier frequency. In fact, under these
hypotheses one can describe a time delay in terms of fractions of the carrier period.
When the PAA is transmitting broadband signals, the approximation above does
not hold anymore: in fact, in this case a constant phase shifting of all the spectral
components in the broad signal spectrum would give different frequencies a
different delay. Therefore, when the signal is broadband, the phase shifting
approach induces a beam squinting: different frequencies of the signal spectrum
aim at a different angle, losing directivity and gain in the antenna. In order to
avoid it, a true time delay (TTD) must be controlled at each antenna element. In
current electronically controlled PAAs, the phase shifting is realized by means of
analog RF PSs. In case of broadband signals, instead, the TTD is implemented in
the digital domain (rather than in the analog RF domain) by processing the
numerical signal at each antenna element in order to synthesize a delay on the
samples. This operation requires a huge digital processing capability, and is
usually reserved for high-performance applications only.
In this perspective, photonics can provide promising solutions to meet the 5G
requirements, potentially reducing also the cost of BFN elements, thanks to
photonic integration [6], [7]. If the RF signal is transferred in the photonic domain,
it is easy to implement either a phase shifting or a true time delay, taking
advantage of the huge bandwidth, frequency flexibility, and EMI insensitivity of
photonics. For both the PS and the TTD, several different solutions are available.
In the following discussions, we will describe these methods considering the beam
forming in transmission, but the same approaches can be used for controlling the
direction of detection in receiving PAAs. Let us consider an RF signal loaded on

a laser by means of a single sideband modulation. To implement a phase shifting
to the optically carried RF signal, it is necessary to shift the optical carrier with
respect to the sideband. This way, once the optical signal is converted back to the
RF domain by a photodiode, the variation in the phase difference between carrier
and sideband is transferred to the RF signal. This can be realized by separating
carrier and sideband in an optical de-interleaver (as the MRR-loaded MZ
interferometer presented above), and shifting the carrier alone, e.g., in a phase
modulator, before recombining the carrier and the sideband (Fig. 4a) [8],[9].
Another possible way for shifting the carrier only, is to use a wavelength specific
phase shifter. For example, a microring resonator (MRR) under particular
hypothesis on the coupling between the waveguide and the ring, realizes an all-
pass filter that is induces a 360° steep phase shift across its resonance wavelength.
If the optical carrier is placed close to the MRR resonance, slightly changing the
reciprocal position of the carrier and the microring will change the phase of the
carrier without affecting the sideband (Fig. 4b) [10],[7]. These approaches can
take advantage of the fast phase control ensured by the photonic techniques, so
that phase tuning time faster than 1 ns can be achieved, regardless of the RF carrier
frequency that can easily be as high as several tens of GHz.
Fig 4.
a) Phase shifting through de-interleaving and phase control; b) phase shifting through a
wavelength-specific phase shifter, as a microring resonator in all-pass configuration.
Amp. Resp.
a) b) 1
Phase Ctrl.
νres.
Phase Ctrl.
Phase Resp.
p
DF
F Mod.
De-Inter.
νc
-p
F ctrl
DF DF
DF DF
νc νc+ fRF νc+ fRF νc fRF

νc νc+ fRF νc νc+ fRF fRF
On the other hand, effective TTD based on photonics has been demonstrated
exploiting several different approaches. The most straightforward one takes
advantage of the low loss of optical waveguides to implement variable delays
through optical path switching (Fig. 5a): the signal is switched among few optical
paths of different length, therefore with different propagation times [11]. A second
type of TTD exploits the wavelength dependence of the laser propagation time
through a medium due to the chromatic dispersion [11]-[14]: considering a
modulated laser transmitted through, e.g., an optical fiber, if the laser wavelength
is changed, the propagation time in the dispersive element changes as well.
Therefore, it is possible to control the delay of an optical signal by controlling its
wavelength. If the RF signal to each antenna element is loaded on an independent
laser, by tuning the lasers it is possible to control the time of emission of the RF
signal of every element in the array. For example, the dispersion-compensating
fiber (DCF) has a dispersion of about -100 ps/nm·km; therefore, if a spool of 1
km is available, tuning the carrier wavelength of 1 nm will produce a variation in
the time of arrival equal to 100 ps. Another relevant method for realizing a TTD
using photonics exploits the so-called slow light effect, i.e., emulating the TTD
by controlling the group delay of the optical sideband only [15]. This approach is
based on the concept that the delay is the wavelength derivative of the phase;
therefore, synthesizing phase variations with tunable steepness in the spectrum, it
is possible to control the delay of the optical signal. An elegant implementation
of this method has exploited the group delay of few cascaded optical micro-rings
[16], controlling their resonance frequency through a phase control in the ring,
and their group delay through the coupling between ring and waveguide (Fig. 5b).
This last example is particularly significant since it has demonstrated a photonic-
integrated beam forming network based on TTD for broadband signals at high
frequency. The optical TTD is therefore an extremely interesting solution for
high-performing beam forming. Nevertheless, its implementation is practically
complex, in particular if tunable TTD is realized in integrated photonics, where
several parameters must be controlled simultaneously at each antenna element.
Therefore, if the target application does not make use of ultra-wideband signals,
an approach based on optical phase shifting can be more convenient, trading off a
limited amount of squinting with a significant simplification in the operational
complexity [9].
Fig 5.
a) Photonics-based TTD based on path switching. b) Principle of wideband TTD based on
group delay (GD) synthesis by MRRs.
b) Controls
f1 f2 f3
a)
Input GD1 GD2 GD3 Output
Switch
Switch
Switch
Switch
Controls
Input Output
L 2L 4L
Total GD
Group Delay
GD1 GD2 GD3
f1 f2 f3 Frequency
2 Photonic technologies for 5G

5G wireless communication technology requires an increase in bandwidth of three
orders of magnitude (>500 Mb s−1) for each user and all objects connected to the
Internet, as the 5G evolution is driven by the growing mobile communication
markets and the development of the Internet of Things (IoT). Therefore, there is
urgent demand for technologies that can meet requirements in terms of bandwidth
and power consumption. Photonics is poised to play an increasingly important
role in ICT, since the fixed high capacity links are largely based on photonic
technologies. At present, optical interconnections in data centres are mainly
between boards that provide the platform on which the electronic components and
optical or electro-optical devices are connected. In the near future, the number of
optical interconnections will increase. As a result, by 2021, the production of
optical interconnections is predicted to be >10 million per year. Photonic devices
need to support ultra- large bandwidth operation, for example, 200 Tb s−1 in a
single fibre and >10 Tb s−1 cm−2 in integrated photonics chips. To achieve this,
the key photonic components, e.g. photodetectors and modulators, need very high
performances in terms of speed (≥25 Gb s−1), footprint (<1 mm2), insertion loss
(<4 dB), manufacturability (>106 pieces per year) and power consumption (<1 pJ
bit−1). The photonic devices needed to meet these requirements are typically based
on LiNbO3, III-V semiconductors such as InGaAsP/InP and semiconductors used
in Silicon photonics. However, for technologies to become widespread, devices
must be mass produced, cost efficient, reproducible, reliable, and compliant with
existing semiconductor processes and environmental regulations. With these

considerations in mind, for large- scale production, Silicon photonics is preferable
to others because the technological processes are the same as those already present
in Si foundries commonly used in the semiconductor industry.
2.1 Silicon
Photonic Integrated Circuits (PICs) have become a commercial reality in a

number of markets, especially in telecom and datacom. PICs enable complex
optical and opto-electronic functions on a very compact footprint with high
reliability. And because of wafer-scale manufacturing, the cost of a PIC can be
significantly lower than with conventional technologies (relying on bulk optical
or other assembly platforms) for the same function.
Silicon photonics is the field that takes advantage of more than 50 years of
massive investment in silicon technology for electronic ICs. It leverages the vast
know-how of the complementary metal-oxide-semiconductor (CMOS) world to
develop PICs in the technologies of existing CMOS fabs. The field of silicon
photonics has been growing at an amazing rate, both scientifically and
industrially. Today more than 15 CMOS fabs (industrial fabs or semi-industrial
R&D fabs) around the world have developed a mature process flow for silicon
photonics. This allows to access a much larger scale of manufacturing with high
volumes and low cost with respect to other technologies, e.g. Indium Phosphide.
Silicon photonics comprises passive and active optical devices and systems based
on high refractive index Silicon or Silicon nitride waveguides which allow high-
density integration of very compact devices including: high-speed phase
modulators, phase shifters, switches, multiplexers, couplers, etc. Silicon
photonics may include also other materials to provide functionalities not allowed
by only Silicon, e.g. Germanium or III-V materials. Germanium is typically used
to integrate photodetectors and electro-absorption modulators as it can be directly
grown on Silicon and be processed in a CMOS line. III-V materials are used to
integrate hybrid lasers or semiconductor optical amplifiers (SOAs) on Silicon, e.g.
by wafer bonding or transfer printing techniques.
Silicon photonics is a key technology to satisfy the demanding challenges of next

5G networks: high bandwidth, low power consumption, small footprint. Silicon
photonics optical interfaces have already reached high volumes for cloud data
centers and have played an important role in mobile networking. As data rates of
commercial optical interface increase from 10Gb/s to 25Gb/s, Silicon photonics
will be even more critical to support high-speed, optical connectivity to/from cell
tower. As an example, today Silicon Photonics supports the 100GbE optical
interfaces with PSM4 and CWDM4 transceiver based on different standards.
However, the 100GbE will soon evolve in 200GbE and 400GbE to increase the
throughput of each module.
As examples of real devices a Silicon Photonics Optical Dispersion Compensator
and an Optical Beam forming Networks are reported.
Example: Silicon Photonics Optical Dispersion Compensator
High speed optical direct detection (DD) interfaces based on DWDM technology
could meet the needs of both emerging Cloud and 5G networks [17], provided
that is possible to extend their distance reach up to 20 km. In this scenario,
dispersion compensating devices based on integrated photonics, are appealing
because they can be embedded in plug-and-play optical transceivers similar to the
current Small Form Factor (SFP) modules. In [18] an integrated optical dispersion
compensation module (ODCM) realized on a Silicon photonic (SiP) platform was
presented. The SiP ODCM is able to receive and separate a polarization
multiplexed (PolMux) signal in the C-band (ITU-T 100Ghz DWDM grid), and
compensates the chromatic dispersion accumulated up to 30km single mode fiber
(SMF) transmission with a 10km granularity.
All-pass microring resonators have been demonstrated to be suitable for the
design of optical dispersion compensators (ODC) [19]. An ODC for the
compensation of the typical chromatic dispersion of a standard G.652 SMF at
1550nm is fabricated. The target SMF length is 10km with a constant-dispersion
pass-band width of 50GHz. In order to obtain a flat group delay in the desired
pass-band an ODC composed of three all-pass microring resonators was designed.
The 10km ODC has been designed, on a SiP platform, defining the phase Φ, the
power coupling ratio k and free-spectral range (FSR) of the three ring resonators
[19]. The ODCM is equipped with a polarization active controller (PAC), based
on a previously reported design [20], able to receive and separate the two
polarizations of a PolMux signal into two independent sections of the module.
The two polarizations are then separately processed by two independent cascades
of three 10km ODCs. Mach-Zehnder interferometer (MZI) based optical switches
allows to select the number of 10km ODCs in which the signal will pass through
on each arm, i.e. the SMF length to be compensated: 10km, 20km or 30km. Each
of the employed all-pass ring resonators is equipped with an electric control in
order to adjust the ODC response. Fig. 6 shows the ODCM schematic, the mask
layout each building block and an optical microscope picture of the fabricated
device.
Fig 6.
(a) Schematic of ODC module. (b) Mask layout of each building block. (c) Optical
microscope picture of the fabricated device..
The fabricated ODCM was used in a transmission experiment to assess the

expected extension of the optical reach. The experimental setup is shown in Fig.
7.
Fig 7.
Experimental setup for the transmission experiment and validation of the proposed ODCM
The light of a CW source is intensity modulated (IM) by a dual-drive Mach-

Zehnder Modulator (MZM) driven by a 50Gb/s NRZ data stream (231-1 PRBS)
generated by a programmable bit pattern generator (BPG). The available bit rate
is doubled exploiting PolMux emulated through a split&delay PolMux emulator,
obtaining a 100 Gb/s PolMux NRZ signal. The signal is then sent to different
spools of G.652 SMF to assess the performance at different fiber lengths. At the
receiver side, the incoming signal is coupled into the SiP ODCM through the dual
polarization grating coupler of the PAC [4], which separates the two polarizations
of the PolMux signal in the two arms of the SiP ODCM.
The two tributary signals at the output of the two arms are sent to an erbium doped
fiber amplifier (EDFA) to compensate for the insertion losses of the chip which
are larger than expected due to accidental fabrication impairments. The tributary
signals are then sent to a DD NRZ Rx consisting of a 30GHz photodiode followed
by a 50Gb/s symbol-by-symbol hard threshold detector. Fig. 8a and 8b show the
eye diagrams for the two received polarizations after propagation over 30km of
SMF and the ODCM. Transmission of the 100Gb/s PolMux NRZ signal over
different SMF lengths, 10km, 20km and 30km was characterized by properly
enabling the optical switches of the ODCM in order to process the signal with
one, two or three 10km ODCs, respectively. To evaluate performance, the optical
power required at the 50Gb/s DD NRZ receiver was measured for a pre-FEC BER
= 10-3, that is a typical reference value for hard-decision FEC with 7% overhead,
suitable for Metro application. First the NRZ performance in back-to-back
conditions was measured, where a BER = 10-3 was obtained with an optical power
at the receiver of -12.5dBm. Then, the same measure was performed with different
spools of standard G.652 fiber together with the ODCM. Fig. 5(c) shows the
obtained pre-FEC sensitivity as a function of the transmission length.
Results show a reasonable power penalty of 2.5dB at 30km, thus confirming the
effectiveness of the ODCM to extend the optical reach of high bit rate IM/DD
transmission in a metropolitan network scenario.
Fig 8.
(a) and (b): eye diagrams for the two received 50Gb/s polarizations after propagation over
30Km of SMF and chromatic dispersion compensation and polarization recovery in the
ODCM. (c): optical power required at the direct detection OOK receiver as a function of the
transmission length, for achieving a BER = 10-3.
Example: Optical Beam forming Networks
Optical phase shifting approach has been chosen, with respect to TTD, to
minimize the impact of the optical solution on the standard radio board,
guaranteeing at the same time, the main requirements for 5G . In fact, the
additional broadening of the pointing beam originated by frequency-dependent
response of a PS element is inversely proportional to the signal carrier frequency
and to the number of array elements [21]. Thus, for small-to-moderate number of
PAA elements (as in the case of cellular networks antennas), the squint-induced
beam broadening can be considered a negligible fraction of the “natural” (i.e., in
absence of modulation signal) beam width even for relatively large signal
bandwidth and pointing angles, if the carrier frequency is sufficiently high. For
instance, the calculated beam broadening due to squint effect for a 1D 8-element
antenna array with a carrier frequency of 30 GHz and a 2 GHz signal bandwidth
is less than 7° (for a natural beam width of 28.5°) for a pointing angle as large as
60°, and it reduces to ~1° (with the natural beam being ~13°) when the pointing
angle is 15°. In several 5G envisioned applications, small-size PAA with
relatively limited scanning angles are expected to prove useful services, such as
dynamic calibrations of the antenna line-of-sight between different stations. In
general, since the 5G scenario will involve high bit rates and the envisaged carrier
frequencies can be well beyond 30GHz, the PS approach might be justified by the
strong reduction in the complexity and cost of the BFN architecture compared
with the more performing TTD approach. The proposed solution aims at replacing
the electrical phase shifter with an optical phase shifter as reported in Fig. 9.
The optical phase shifter includes to different blocks: the first one is shared among
all the PAA elements and does not require photonic integration, while the second
one is specific for each PAA element and should be integrated to reduce the
footprint and the power consumption.
In the first block a continuous-wave laser (1558 nm) is modulated by a RF
oscillator (13GHz) in a dual-nested Mach-Zehnder Modulator (DN-MZM), to
obtain a single sideband optical signal. After amplification by an erbium-doped
fibre amplifier (EDFA), the optical signal is split to the PSs in the BFN. The
second block comprises an optical deinterleaver filter (ODF), an optical phase
shifter (OPS), an optical coupler (OC), and a photodiode (PD). The operation of
this clock relies on optical carrier-sideband separation performed by the ODF and
subsequent phase shift of the isolated carrier in the OPS stage, before the two
components are recombined in the OC and sent to the PD to generate the phase-
shifted microwave signal [22].
Fig 9.
OBFN architecture
Data
IF/RF Phase-controlled
Clock data
Optical
LO
Phase shifter X RF/IF
Mixer
Φ CRTL
LASER
n
PHOTODIODE
COUPLER
φ
FILTER
LO SSB-MOD TO THE
MIXER
fLO
N replicas for an arrayed antenna with N elements

PIC
n 1:N
…
A single circuit shared among the

N antenna elements RF/Electrical Signal
Optical Signal
OPTICAL BOX
The phase of the down-converted signal at the PD output generated by the beating
between the carrier and the sideband components is indeed given by the phase
difference of the two beating signals. Assuming a perfect carrier-sideband
isolation at the ODF outputs, the amount of optical phase shift φ experienced by
the optical carrier in the OPS translates into an equal phase shift φ of the
microwave signal at fRF without any amplitude modulation. The PSs are all
calibrated to centre the filtering port of the DI to the laser wavelength.
This second scheme has been realized in SOI technology which simultaneously
matches the demand for wide phase-shift range, broad bandwidth, low in-band
power oscillations, fast (i.e., sub-μs) reconfiguration speed, and compatibility
with CMOS integrated circuits fabrication technology. The phase modulator is
realized as a reverse-biased Si p-n junction, while the ODF is implemented by a
micro ring resonator (MRR)-loaded Mach-Zehnder interferometer structure.
The photonic integrated circuit (PIC) is able to perform stable phase shifts well in
excess of 360° over a bandwidth of 6 GHz for RF carriers spanning in the X, Ku,
an mm-wave bands, with limited in-band power variations of ~1 dB, and small
response time below 1 ns. A picture of one PS element composed of the PIC and
its driving circuitry is shown in Fig. 10.
Fig 10.
Picture of the photonics-based PS, with the PCB (printed circuit board) hosting the PIC.
Inset: Microscope picture of the PIC
The photonics-based BFN has been characterized for a 4-element PAA. We

analysed the static and dynamic behavior of the RF beam controlled by the BFN.
The main specifications are reported in Table 1. The OBFN works at 13 GHz and
all odd multiples, due to the periodical behavior of the ODF, with a bandwidth
>400 MHz. The use of photonic integration introduce high losses (10 dB),
reducing the power efficiency. However, the faster (<1ns compared to the state of
the art value > s) and more precise (<1# compared to the state of the art value >
4°) beam steering guaranteed by the photonics-based phase shifters makes this
approach a promising solution to BF for PAAs in future 5G networks.
RF frequency 13 GHz and odd multiples

Phase precision 0.06 deg/mV
Response time <1ns
3-dB bandwidth 400MHz
PIC loss (including 4dB for
10 dB
coupling)
Total RF loss 40dBm
Output RF power -40dBm
Table 1: OBFN specifications
For the static characterization, power and phase responses of each PS element are
measured on a vector network analyser while gradually changing the DC voltage
applied to the phase modulator in the PIC. The results are reported in Fig. 11. Fig.
11a shows the phase shift produced on the 13 GHz output RF signal by each PS.
The 4 curves exhibit a good linearity, with a phase variation in excess of 360°
over less than 7 V. Fig. 11b shows the power fluctuations induced on the RF signal
by varying the control voltage: a power variation within ~4 dB is measured over
a 360° phase shift. In fact, a change in the polarization of the p-n junction induces
a change in its refractive index, i.e. on its transparency. This entails a variation of
the attenuation of the optical signal propagating through the phase modulator;
hence, the RF power variation. In the following power variations effect on the
beam emitted by a PAA is considered.
Fig 11.
Characterization of the phase shift (a) and output power (b) of the 4 PSs, versus the applied
voltage to the optical phase modulator.
The radiation pattern of a PAA can be obtained by multiplying the radiation

pattern of the single element with the array factor AF(θ) [23]
Considering four antenna elements spaced by λ/2, , the AF diagram of an ideal
PAA pointing at 0° can be calculated as reported by the dashed red line in Fig.
12a. Instead, the AF of the photonics-based BFN is calculated by measuring on
the oscilloscope the relative amplitude an and actual phase shifts Δθn of each PS,
for a steering angle θ0 = 0°. The obtained AF is represented by the solid black line
of Fig. 12a). As expected, the ideal and real AFs coincide. Then, the phase settings
are changed for pointing at θ0 = -60°. The calculated AF is reported in Fig. 12b)
compared with the ideal one. The curves for the ideal and real AF are in good
agreement, both showing the maximum at -60°, but a difference in the shape can
be recognized. This is due to the optical power fluctuations in the PSs studied in
Fig. 12b). A modification of the PIC is already under study to suppress these
fluctuations and hence improve the beam forming accuracy of the BFN. In the
dynamic characterisation, the PSs are driven to simultaneously switch between
the phase settings θ0 = 0° and θ0 = -60°; their outputs are acquired by a real-time
oscilloscope and processed. The red solid curve in Fig. 12c) shows the RF power
behaviour seen from the view angle θ = 0° during the steering from 0° to -60°,
while the blue dashed curve reports the RF power observed at the same time from
the view angle θ = -60°. The power behaviour clearly describes the effective
steering of the PAA, changing quickly from a static condition at θ0 = 0° to a static
condition at θ0 = 60°. The power levels seen from θ = 0° and 60° are not equal,
as it would be expected, due to the power fluctuations discussed above. The time
needed by the BFN to switch the beam between the two angles is about 5 ns, in
line with the requisites for reconfiguring PAAs in 5G networks. The switching
time of the bare PIC was previously estimated in ~1 ns. As a fact, the measured 5
ns is limited by the bandwidth of the available step-function generator (80 MHz)
used for controlling the PSs, confirming the estimation of the previous analyses.
Fig 11.
Comparison of ideal (dashed red lines) and real (solid, black lines) array factor diagrams for
a PAA pointing at 0° (a) and at -60° (b). (c): Power emitted by the PAA as seen from view
angles at 0° and -60° during the steering switch.
2.2 Hybrid integration and innovative technologies
Silicon photonics enables the integration of many optical functions on a chip

allowing compactness, low cost and large scale of manufacturing. However,
Silicon is an indirect bandgap semiconductor, i.e. it is not suitable for optical
amplification and lasing. Today the most straightforward method to integrate laser
sources with Silicon photonics is co-packaging of separate chips, i.e. the Silicon
photonics PIC and the laser, inside a single package, as shown in Fig. 12. This
approach has the advantage to exploit at maximum the different technologies, but
at the same time introduces some drawback, e.g. in terms of compactness and
extra losses due to optical interfacing of the two worlds.
Fig. 12
a) Co-packaged multi-chip: Silicon PIC, silicon EIC and laser [23]. b) Wafer bonded laser [24].
The integration of lasers and optical amplifiers is also possible by heterogeneous

integration of Silicon photonics and III-V semiconductor technologies. The
hybrid integration of III-V semiconductor optoelectronic components onto Silicon
photonics PICs can be obtained in different ways, as for example flip-chip
integration, hetero-epitaxial growth, wafer bonding or transfer printing.
Flip-chip integration has the advantage that the devices are fabricated on their III-
V native substrate and then glued on the Silicon photonics PIC, which requires
accurate alignment in the assembly process. The hetero-epitaxial growth approach
allows for front-end, wafer-scale integration of the III-V materials, however it is
quite difficult to obtain high quality materials. Wafer bonding approaches
combine some of the advantages of flip-chip integration because the III-V
material is grown on its native substrate and then bonded on Silicon, with the
manufacturability of wafer-scale integration as the device fabrication is finalized
on the Silicon photonics process flow. The transfer printing approach is an
alternative bonding technology offering some advantages. The devices are
partially processed on the native substrates and then transferred using a PDMS
stamp to the target Silicon photonic PIC.
Hybrid integration of laser sources with Silicon photonics is currently developed
in commercial 100GbE Silicon photonics modules ready to support next 5G
wireless Radio Access Network (RAN) fronthaul. Hybrid integration of
semiconductor optical amplifiers (SOAs) is also very important to develop low
loss optical networks. A lot of effort is spent worldwide for the integration of III-
V SOAs on complex Silicon Photonic switching matrix.
At present, among the established technologies the most promising is Silicon
Photonics. However, to date, the requirements of the next 5G network seem to be
not fulfilled in just one system. New materials have emerged to expand the
functionalities of current technologies with better performance in terms of
efficiency and cost. Graphene photonics is emerging as an innovative solution
fully compatible with Silicon Photonics and able to improve the performance of
current Silicon photonics devices.
Graphene is an allotrope of Carbon in which Carbon atoms are arranged in a single
atom thick hexagonal lattice, i.e. a 2D material. Graphene exhibits peculiar
optoelectronic properties and can be used for light modulation and photodetection.
Graphene-based integrated photonics for high-speed datacom and telecom
applications have been demonstrated [24]. Graphene-based photodetectors
integrated on Silicon Photonics already reached good responsivities and operation
bandwidth Graphene-based modulators have been demonstrated to provide
advantages over Si- based modulators. They are capable of broadband electro-
absorption and electro-refraction operation with improved efficiency with respect
to the standard Silicon Photonics platform.
Graphene devices are compatible with complementary metal oxide semiconductor
(CMOS) processing, and enable post-processing fabrication and the use of
different substrates. Graphene devices have also the advantage that does not
require implantations on Silicon or Germanium hetero-epitaxy. Hence, the
waveguide can be Silicon, Silicon Nitride, Silica or another transparent material.
Practically, this implies a post-processing shift in manufacturing from front-end
to back-end-of-line. In addition, Graphene technology does not necessarily
require expensive SOI wafers, or implantation for junctions, and Ge growth for
detectors. Because SiN and SiO2 waveguides are wider than Si photonics ones,
the lithography node can be relaxed. The waveguide size is ~0.5 μm for Si, ~1.5
μm for SiN and ~ 8μm for SiO2. All these factors will simplify the technology
and reduce costs.
References
1. NGMN, “5G White Paper,” Feb. 2015 [Online]. Available:
https://www.ngmn.org/uploads/media/NGMN_5G_White_Paper_V1_0.pdf.
2. [2] Ericsson, “Ericsson Mobility Report,” Nov. 2015 [Online]. Available:
http://www.ericsson.com/res/docs/2015/mobilityreport/ ericsson‑mobility‑report‑nov‑
2015.pdf.
3. Ericsson, “Cloud RAN,” White Paper, Sept. 2015 [Online]. Available:
https://www.ericsson.com/res/docs/whitepapers/wp‑cloud‑ran.pdf.
4. J. Beas, G. Castanon, I. Aldaya, A. Aragon-Zavala, G. Campuzano, “Millimeter-wave

Frequency Radio over Fiber Systems: A Survey”, IEEE Commun. Surveys Tuts., v.15, n.4,
2013
5. A. Osseiran, et al., “5G Mobile and Wireless Communications Technology”, Cambridge
University Press, June 2016.
6. L. Zhuang, et al., “Novel ring resonator-based integrated photonic beamformer for broadband
phased array receive antennas-Part II: Experimental prototype”, Journ. Lightwave Technol.,
28 (1), 2010, pp. 19-31.
7. D.B. Adams, and C.K. Madsen, “A Novel Broadband Photonic RF Phase Shifter”, J.
Lightwave Technol., 26 (15), 2008, pp. 2712-2717.
8. G. Serafino, C. Porzi, V. Sorianello, P. Ghelfi, A. D’Errico, S. Pinna, M. Puleri, M.
Romagnoli, A. Bogoni, “Design and Characterization of a Photonic Integrated Circuit for
Beam Forming in 5G Wireless Networks”, MWP 2017.
9. F. Falconi, C. Porzi, S. Pinna, V. Sorianello, G. Serafino, M. Puleri, A. D’Errico, M.
Romagnoli, A. Bogoni, P. Ghelfi, “Fast and Linear Photonic Integrated Microwave Phase-
Shifter for 5G Beam-Steering Applications”, OFC 2018.
10. M. Pu, L. Liu, W. Xue, Y. Ding, H. Ou, K. Yvind, and J. M. Hvam, “Widely tunable
microwave phase shifter based on silicon-on-insulator dual-microring resonator”, Optics
express, 18(6), 2010, pp. 6172-6182.
11. R. Soref, “Optical dispersion technique for time-delay beam steering”, Applied Optics, vol.
31, no. 35, pp. 7395-7397, 1992.
12. L. Yaron, R. Rotman, S. Zach, M. Tur, “Photonic Beamformer Receiver With Multiple Beam
Capabilities”, IEEE Photonics Technology Letters, vol. 22, no. 23, pp. 1723-1725, 2010.
13. K. Prince, M. Presi, A. Chiuchiarelli, I. Cerutti, G. Contestabile, I.T. Monroy, E. Ciaramella,
“Variable Delay With Directly-Modulated R-SOA and Optical Filters for Adaptive Antenna
Radio-Fiber Access”, IEEE Journal of Lightwave Technology., vol 27, no. 22, pp. 5056-
5064, 2009.
14. F. Scotti, P. Ghelfi, F. Laghezza, G. Serafino, S. Pinna, A. Bogoni, "Flexible True-Time-
Delay Beamforming in a Photonics-Based RF Broadband Signals Generator," IET
Conference Proceedings, pp. 789-791, 2013.
15. A. Zadok, O. Raz, A. Eyal, M. Tur, “Optically Controlled Low-Distortion Delay of GHz-
Wide Radio-Frequency Signals Using Slow Light in Fibers”, IEEE Photonics Technology
Letters, vol. 19, no. 7, pp. 462-464, 2007.
16. A. Meijerink, C.G. Roeloffzen, R. Meijerink, L. Zhuang, D.A. Marpaung, M.J. Bentum, M.
Burla, J. Verpoorte, P. Jorna, A. Hulzinga, W. Van Etten, “Novel Ring Resonator-Based
Integrated Photonic Beamformer for Broadband Phased Array Receive Antennas—Part I:
Design and Performance Analysis”, IEEE Journal of Lightwave Technology, vol. 28, no. 1,
pp. 3-18, 2010.
17. X. Liu et al., "Emerging optical access network technologies for 5G wireless," IEEE/OSA
Journal of Optical Communications and Networking, vol. 8, no. 12, pp. B70-B79 (2016).
18. V. Sorianello, G. De Angelis, F. Fresi, F. Cavaliere, L. Potì, M. Midrio, and M. Romagnoli,
"100Gb/s PolMux-NRZ Transmission at 1550nm over 30km Single Mode Fiber Enabled by a
Silicon Photonics Optical Dispersion Compensator," in Optical Fiber Communication
Conference, OSA Technical Digest (online) (Optical Society of America, 2018), paper
W2A.31.
19. C. K. Madsen et al., "Integrated all-pass filters for tunable dispersion and dispersion slope
compensation," IEEE Photonics Technology Letters, vol. 11, no. 12, pp. 1623-1625 (1999).
20. P. Velha et al., “Wide-band polarization controller for Si photonic integrated circuits”, Opt
Lett. 41(24), pp. 5656-5659 (2016).
21. Hansen, Robert C. Phased array antennas. Vol. 213. John Wiley & Sons, 2009.
22. V.J. Urick et al., “Microwave phase shifting using coherent photonic integrated circuits”,
IEEE J. Sel. Topics Quant. El., vol. 22, no. 6, Dec. 2016.
23. F. Amato et al.,“Ultra-fast Beam Steering of a Phased-Array Antenna Based on Packaged
Photonic Integrated Circuits” in Proc. ECOC2018, Rome, Italy, Spet. 2018.
24. http://www.luxtera.com/silicon-photonics-technology/
25. Intel. A Hybrid Silicon Laser Silicon photonics technology for future tera-scale computing,
White Paper. (2006)
26. M. Romagnoli, et al., Graphene-based integrated photonics for next-generation datacom and
telecom, Nat. Rev. Materials vol. 3,pp. 392-414 (2018)
Impulse Radio NOMA Communications in 5G
and Beyond1
Mai T. P. Le, Giuseppe Caso, Luca De Nardis, Maria-Gabriella Di Benedetto
Abstract Motivated by recent theoretical challenges for 5G and beyond 5G systems,

this chapter aims to position relevant results in the literature on code-domain non-
orthogonal multiple access (NOMA) from an information-theoretic perspective,
given that most of the recent intuition of NOMA relies on another domain, that is, the
power domain. Theoretical derivations for several code-domain NOMA schemes are
reported and interpreted, adopting a unified framework that focuses on the analysis
of the NOMA spreading matrix, in terms of load, sparsity, and regularity features.
The comparative analysis shows that it is beneficial to adopt extreme low-dense code-
domain NOMA in the large system limit, where the number of resource elements and
number of users grow unboundedly while their ratio, called load, is kept constant.
Particularly, when optimum receivers are used, the adoption of a regular low-dense
spreading matrix is beneficial to the system achievable rates, which are higher than
those obtained with either irregular low-dense or dense formats, for any value of load.
For linear receivers, which are more favorable in practice due to lower complexity,
the regular low-dense NOMA still has better performance in the underloaded regime
(load < 1), while the irregular counterpart outperforms all the other schemes in the
overloaded scenario (load > 1).
1 Introduction
Being acknowledged as an important enabler for 5G multiple access, non-orthogonal

multiple access (NOMA) with its diverse dialects recently attracted a huge attention
Mai T. P. Le, Giuseppe Caso, Luca De Nardis, Maria-Gabriella Di Benedetto

Department of Information Engineering, Electronics and Telecommunications, Sapienza University
of Rome, Rome 00184, Italy.e-mail: mai.le.it@ieee.org
1 This
chapter is an extract version with some modifications of the paper [1] published in
IET Communications and is subject to Institution of Engineering and Technology Copyright.
The copy of record is available at the IET Digital Library.
277
278 Mai T. P. Le, Giuseppe Caso, Luca De Nardis, Maria-Gabriella Di Benedetto
in the wireless community from both industry and academia [2]. In fact, NOMA
has been currently proposed by the 3rd Generation Partnership Project (3GPP) for
5G New Radio (NR). Many NOMA schemes for NR were initially proposed in
Release-13 (Rel-13) Study Item, and were partly analyzed in Rel-14. While some
specifications of 5G-NR have been officially standardized in the current Rel-15 (for
example, Specification TS 38.211 for Physical channels and modulation [3]), Spec-
ification TS 38.812 for Study on NOMA for NR [3] is still under investigation with
the expectation to have a ‘ready’ NR system in 2020 [4]. Furthermore, NOMA is a
strong candidate for beyond 5G systems, thanks to its capability of supporting mas-
sive communications [5]. In traditional orthogonal multiple access (OMA) schemes,
users are allocated to orthogonal resource elements (REs) either in time, frequency,
or code domains [6]. Based on the relationship between the number of users K and
the number of REs N, also called degrees of freedom, the system is termed under-
loaded when K < N and overloaded when K > N. As a matter of fact, one should
note that underloaded systems can be made OMA under the assumptions of perfect
knowledge of channel state information and perfect synchronization between the
transmitters and receivers. A very different situation arises when the system is over-
loaded. Even under ideal conditions such as ideal propagation and ideal allocation
strategy, the system is intrinsically affected by ‘collisions’ due to interference [7].
This scenario may be easily envisioned in 5G, for example in the internet-of-things
(IoT), in which a huge number of terminals are required to transmit simultaneously.
Noteworthy that, according to this understanding, all conventional OMA schemes,
including the well-known CDMA, become NOMA in the overloaded regime, due to
the exceedingly large number of users compared to the number of REs.
In order to enable detection at the receiver side, different users are detected based
on the difference of power or spreading codes, leading to two main corresponding
approaches: power-domain NOMA vs. code-domain NOMA. To address interference
provoked by the lack of a sufficient number of REs, controllable interference among
REs may be introduced in the code domain with an acceptable complexity of receivers
[8]. This NOMA approach is currently known as code-domain NOMA [6]. However,
with reference to NOMA, most of recent works in the literature focus on the power-
domain case [2], which is based on the idea of serving multiple users at the same
time/frequency/code with different power levels [2, 6, 9]. This chapter, on the other
hand, makes an effort to contribute to the understanding of code-domain approach,
particularly from an information-theoretic perspective.
In the context of massive connectivity, expected for 5G and beyond 5G systems,
the number of users is supposed to be very large compared to the number of REs.
The behavior of the system should thus be considered in the asymptotic limit, where
both K and N go to infinity, while the ratio K/N = β, called load, remains finite.
This corresponds to analyzing the system in the large system limit (LSL) [10].
NOMA schemes and corresponding achievable rates can be investigated in the LSL
considering three main features:
• load β, particularly β > 1, also known as overloading factor, which is considered
as a significant feature of code-domain NOMA [6].
Impulse Radio NOMA Communications in 5G and Beyond1 279
• sparsity, describing the structure of code-domain NOMA spreading matrix, whose

spreading sequence is commonly designed to be sparse to reduce the detection
complexity [6].
• regularity, which characterizes possible spreading mapping constraints [11]. As
will be detailed in Sec. 3, regular low-dense NOMA refers to the case where the
number of users per occupied RE and the number of occupied REs per user are
fixed, whereas irregular designates the case where these numbers are random and
fixed on average [11, 12, 13].
This chapter addresses a general code-domain NOMA analysis by mapping rel-
evant information-theoretic results in the literature on code-domain NOMA, from
which the corresponding information-theoretic results are expected to explore the
relationship between the achievable rates and the aforementioned peculiar features
of code-domain NOMA. Understanding the behavior of the system in terms of
information-theoretic bounds can provide crucial insight to select system param-
eters, and can contribute as a reference for future release of 5G standardization
[6].
2 A reference mathematical model for Code-domain NOMA
Apart from the fact non-orthogonality feature of NOMA has been used recently [2],
initial NOMA concept has a long story from the beginning of 1990s. Non orthogonal
signal sets with particular structures were invented such that they may be detectable at
the receiver. First investigations on designing guidelines for non-orthogonal spread-
ing codes were made by Ross and Taylor [14, 15] that were applicable to an over-
loading system. By adding additional linearly dependent codes while maintaining
the orthogonal minimum distance (Euclidean distance) to ensure feasible detection,
the signal sets essentially became non-orthogonal. In addition, these sets were put
under the constraint as such all users should not have higher powers than that of
orthogonal set. On the receiver side, iterative decoding algorithm such as message
passing algorithm (MPA) detector, also known as belief propagation (BP) algorithm
in low-density parity-check (LDPC) codes, is employed.
CDM-NOMA exploits code to distinguish different users at the receiver, i.e. it
function similarly to the traditional DS-CDMA system. The main feature identifying
CDM-NOMA from CDMA is by employing sparsity in spreading sequences via
low-density or low cross-correlation sequences. Based on the specific scenario,
single-carrier or multi-carrier NOMA schemes can be adopted, corresponding to
single-carrier DS-CDMA or multi-carrier CDMA (MC-CDMA), respectively. The
15 existing proposals of NOMA proposed for the Rel-14 3GPP NR Study item [3]
(Table 1) and recent CDM-NOMA proposals available in the literature, therefore,
will be classified based on dense vs. low-dense and single vs. multi carrier features.
In addition to proposed methods for NOMA from Table 1, including MUSA
[16], PDMA [17, 18], IGMA [19], IDMA [20, 21], other NOMA schemes such as
LDS-CDMA [22] (also known as time-hopping (TH-CDMA) [23, 24]), SAMA [25]
NOMA schemes Full Name Company UL/DL

1 Power-domain NOMA Power-domain non-orthogonal multiple access DCM UL/DL
2 SCMA Sparse code multiple access Huawei UL/DL
3 MUSA Multi-user shared access ZTE UL/DL
4 PDMA Pattern division multiple access CATT UL/DL
Low code rate and

5 LSSA ETRI UL
signature based shared access
6 RSMA Resource spread multiple access Qualcomm UL
7 IGMA Interleave-grid multiple access Samsung UL/DL
8 IDMA Interleave division multiple access Nokia UL
9 NCMA Non-orthogonal coded multiple access LGE UL
10 NOCA Non-orthogonal coded access Nokia UL
11 GOCA Group orthogonal coded access MTK UL
Low density spreading -

12 LDS-SVE Fujitsu UL/DL
signature vector extension
13 FDS Frequency domain spreading Intel UL
14 LCRS Low code rate spreading Intel UL
15 RDMA Repetition division multiple access MTK UL
Table 1: NOMA schemes proposed for the Rel-14 3GPP NR Study Item [3]
will be shown tightly correlated to traditional single-carrier DS-CDMA, therefore,

are classified as single-carrier NOMA. By contrast, LDS-OFDM [26, 27], along
with remained NOMA schemes in Table 1 including SCMA [28, 29], LSSA [30],
NCMA [31], NOCA [32], GOCA [33], LDS-SVE [34], LCRS/FDS [35], RDMA
[33] are well fitted to multi-carrier CDMA model, hence, classified here as multi-
carrier NOMA. RSMA [36, 37] is a NOMA dialect that is proposed for both type of
waveforms, depending on the specific application scenario.
Code-domain NOMA classification is illustrated in Fig. 1.
Code-domain
NOMA
Single-carrier Multi-carrier
Dense Low-dense Dense Low-dense
DS-CDMA LDS-CDMA MC-CDMA LDS-OFDM

IDMA SAMA MC-RSMA SCMA
SC-RSMA MUSA LCRS/FDS NCMA
PDMA RDMA NOCA
IGMA GOCA LDS-SVE
Fig. 1: Code-domain NOMA classfification
2.1 Single-carrier NOMA
For single-carrier NOMA, each data symbol of user k is spread by N chips of

the corresponding spreading sequence, that is similar to as DS-CDMA system.
Naturally, the mathematical model of single-carrier NOMA may be built from the
baseline model of DS-CDMA, proposed by Verdú and Shamai in [10] and [38] as
follows
y = SHb + n, (1)
where the received signal y ∈ C N belongs to a space characterized by N signal
dimensions. Note that N also represents the number of elements over which each
symbol is spread, that is the number of REs, and equivalent to the number of
‘chips’, as termed commonly in CDMA. Vector b = [b1, . . . , bK ]T ∈ CK is the
vector of symbols transmitted by K users. Being a random spreading matrix, S =
[s1, . . . , sK ] ∈ C N ×K is composed of K columns, each being the spreading sequence
sk of user k (1 6 k 6 K). Supposing the channel is flat, the channel matrix can be
represented as H = diag [h1, . . . , hK ] ∈ CK×K , whereas it reduces to the identity
matrix if the AWGN channel is assumed. Lastly, the noise n ∈ C N is described by a
circularly-symmetric Gaussian vector with zero mean and covariance N0 I.
The nature of the representation matrix S defines the multiple access methods.
This matrix is known as spreading matrix or signature matrix in DS-CDMA, TH-
CDMA, LDS-CDMA, SAMA, or code matrix in MUSA, and pattern matrix in
PDMA. As a matter of fact, NOMA schemes can be classified into dense vs. low-
dense, where the corresponding matrix S is dense if all REs are used vs. low-dense,
when some REs are not used. In terms of energy, this corresponds to having all REs
contain signal energy vs. energy is concentrated on only part of the available REs,
reflected by the presence of nonzero entries in S. According to this understanding,
DS-CDMA inherently stands for dense spreading. For single-carrier NOMA, the
dense group includes DS-CDMA and IDMA, while the low-dense group includes
LDS-CDMA, TH-CDMA, SAMA, MUSA, IGMA and PDMA.
2.2 Multi-carrier NOMA
For multi-carrier NOMA schemes, the system can be described as a combination

of CDMA for dense NOMA (respective, LDS-CDMA for low-dense NOMA) with
multi-carrier modulation, for e.g. OFDM, that makes multi-carrier NOMA work
analogously to MC-CDMA [39]. Based on this intuition, the system model of multi-
carrier NOMA basically can be elaborated as below, assuming the number of OFDM
subcarriers shared by every user is also equal to the spreading gain N [40].
In multi-carrier NOMA, each data symbol of user k is replicated into N parallel
copies, each copy is then multiplied by a chip from the respective spreading sequence.
All N copies are then mapped to N subcarriers and are transmitted in parallel. Thanks
to inverse discrete Fourrier transform (IDFT) implementation, those N parallel chips
are converted into serial sequence for further transmission [40].
Adopting the same notation of spreading matrix S = [s1, . . . , s K ] with sk being
frequency-domain spreading sequence of user k as in eq. (1), the baseband signal by
the kth user in time-domain is expressed by
W s k bk .
Here W denotes the N × N IDFT matrix, and bk again stands for the data symbol of
the kth user.
As a matter of fact, the receiver consists of N matched filters, corresponding to N
subcarriers, is equivalently to conducting a discrete Fourrier transform (DFT) on the
discrete baseband domain. Therefore, the received vector by user k in the frequency
domain is bk s̃k , where
s̃k = diag[h11, . . . , hkN ]sk ,
with hki being the fading coefficient at subcarrier i of the kth user.
Since in practice, each subcarrier is narrow enough to experience only flat fading,
the system model of multi-carrier NOMA is, therefore, mathematically equivalent
to that of single-carrier NOMA in (1) with respect to flat-frequency fading. This
observation is also reported by Tulino et al. [40] for the case of DS-CDMA and MC-
CDMA. Here the spreading matrix S ∈ C N ×K again defines different multi-carrier
NOMA schemes.
3 Theoretical analysis of Code-domain NOMA
In this section, the theoretical behavior of NOMA is analyzed under the impact
of three main factors, that are: load, sparsity and regularity. First, the load factor
provides a straightforward way to study the system behavior in the underloaded
(β < 1) vs. overloaded (β > 1) regimes. As mentioned above, overloaded systems
are necessarily NOMA, since as soon as β overcomes the boundary value β = 1,
new users find all REs occupied.
The second feature, sparsity, evaluates system from dense to extreme low-dense
based on Ns , the number of used dimensions, defines as the degree of sparseness.
If Ns = N, the system is dense. If Ns = 1, then the system is extreme low-dense.
All other degrees of sparseness lie in between these two extreme cases. The dense
vs. low-dense feature is directly reflected by the properties of matrix S, where ‘0’ in
the matrix indicates elements with zero energy. A heuristic way to think of NOMA
scheme is thus as a version of the overloaded CDMA scheme and low-dense NOMA
can be referred to as sparse overloaded CDMA [41]. Naturally, it is expected to
investigate the effect of those NOMA parameters, including the load β and the
degree of sparseness Ns , on theoretical behavior of dense vs. low-dense NOMA.
Achievable rates of low-dense NOMA in the LSL were early and extensively
evaluated via sparse CDMA by means of the replica method, also known as heuristic
statistical physics, in [41, 42, 11, 43]. Since the derivations provided by replica
method were typically non-rigorous, the information-theoretic analyses on low-
dense NOMA were found rigorously via closed-form expressions in LDS/TH-CDMA
model [23, 24], and in regular sparse NOMA model [12, 13]. Given that multiple
system models are possibly proposed due to different assumptions, below we reported
all curves along with the existing relevant theoretical results in our unified model
(c.f. 2).
Regarding the third feature, the regularity, low-dense NOMA (Ns < N) are further
classified into irregular vs. regular based on spreading mapping constraints, given
that Ns is also the number of occupied REs per user, whereas N is the total number
of REs per user. Previous works on sparse CDMA and low-dense NOMA were
classified as irregular since the number of occupied REs per user was randomly
Poissonian distributed with fixed mean [41, 42], and randomly uniformly distributed
[23, 24]1, respectively. On the other hand, in terms of spreading matrix, the regularity
assumption in [12, 13] requires matrix S be structured with exactly Ns ∈ N+ and
βNs ∈ N+ non-zero entries per column and row, respectively. It is equivalent to have
each user occupying Ns REs and each RE being allocated with exact βNs users,
subject to Ns and βNs being integers. It is, in general, challenging to have such
an ideal model in practical scenarios where users are not allowed to independently
select the spreading sequences, they must be coordinated or central scheduled [12].
The regular low-dense NOMA via regular sparse CDMA was early demonstrated
to be superior to the dense in terms of bit error rate in high noise regime in [11],
1 The irregular low-dense NOMA in [23, 24] is called as partly-regular sparse NOMA in [12, 13].
and in terms of spectral efficiency via explicit analytical expressions in recent works
[12, 13].
In the following, theoretical behavior of irregular vs. regular low-dense NOMA
(Ns < N) is analyzed with the adopted reference models LDS/TH-CDMA [23, 24]
vs. regular sparse NOMA [12, 13], respectively. DS-CDMA is adopted as a represen-
tative of the dense NOMA group (Ns = N) [38]. Both optimal and linear receivers
are considered in all cases. Spectral efficiency expressions [bits/s/Hz] for different
cases are reported for the self-contained purpose of the chapter. It is important to
notice that the theoretical results of irregular low-dense NOMA are available only
for Ns = 1 [23, 24], while closed-form expressions of the regular case are valid
only for intermediate degrees of sparseness, specifically for Ns > 2, βNs ∈ N+ [13].
For irregular low-dense NOMA, since the closed-form expressions for intermediate
Ns do not exist yet in the literature (and in general, are not easy to achieve), the
results will be shown via Monte Carlo simulations for a full coherent overview. For
regular low-dense NOMA, the regularity in case of Ns = 1 yields a typical setting,
which includes a set of parallel Gaussian multiple access channels (MAC), that has
been recently investigated in [1]. The references for mapping information-theoretic
results in code-domain NOMA are summarized in Table 2, with the corresponding
literature.
Table 2: Summary of available theoretical bounds with corresponding references

in the literature
Dense NOMA Low-dense NOMA

(Ns = N ) (1 < Ns < N ) (Ns = 1)
Irregular Regular Irregular Regular
Literature [38] [23] [13] [23] [1]
3.1 Dense vs. Irregular low-dense NOMA
In this part, theoretical behavior of dense vs. irregular low-dense NOMA is analyzed
with the two corresponding reference models, that are DS-CDMA (Ns = N) [38] and
LDS/TH-CDMA (Ns < N) [23]. Since the AWGN channel is used for both cases,
the channel matrix H in eq. (1) becomes an identity matrix. The only difference
in the mathematical model between DS-CDMA and LDS-CDMA is situated in the
sparseness of matrix S. In DS-CDMA (Ns = N), all entries of S are randomly filled
by binary values of {±1}, while in LDS-CDMA, for example with Ns = 1, each
column of S, representing a user, contains only one nonzero entry ({+1} or {−1}),
and all the rest are nil.
Fig. 2: Achievable rates (bits/s/Hz) of dense NOMA vs. irregular low-dense NOMA as a function
of β with fixed Eb /N0 = 10 [dB] (parts of the data used to draw this figure were extracted
from [23])
Figure 2 shows the achievable rates of dense vs. irregular low-dense systems with
optimum and linear receivers as a function of β with fixed value of Eb /N0 = 10
[dB]. With respect to load factor β, the border line (vertical dashed line) at β = 1
divides Fig. 2 into two areas corresponding to OMA (underloaded with β < 1) and
NOMA (overloaded with β > 1), with dark and light shaded area, respectively.
In the LSL, Fig. 2 shows that for optimum receivers, dense systems always
outperform irregular low-dense, irrespective of β, that is, whether OMA or NOMA.
Achievable rates for the irregular type drop with Ns from the dense case (Ns = N) to
the extreme low-dense case (Ns = 1), and the gap between the irregular low-dense
and dense becomes negligible at Ns = 2, and tends to vanish from Ns > 2, e.g.
Ns = 5. On the other hand, the behavior of linear detection changes, with respect to
the level of density of the system. For MMSE receivers, achievable rates of the dense
systems are higher than the irregular low-dense in the OMA area, while this situation
is inverse in the NOMA area, starting from about β > 1.2. With growing Ns , for
example Ns = {2, 5}, the gap between the achievable rates of irregular low-dense
NOMA with the SUMF receiver and dense NOMA, sharply reduces, to converge to
the SUMF dense curve. Given that optimum detection is unfeasible to implement in
practice due to the receiver complexity, the above observation provides the ground
for suggesting irregular low-dense NOMA in the LSL, for example, for irregular
low-dense case with Ns = 1.
The reported analysis holds in the case of flat-fading channel, as investigated and
proved in [24].
3.2 Dense vs. Regular low-dense NOMA
Theoretical analysis of regular low-dense NOMA is investigated in this part, in

comparison with the dense case.
• 1 < Ns < N: Closed-form expressions of regular low-dense NOMA achievable

rates for both optimum and MMSE receivers in [13] are valid under the following
constraints:
– if each user has 2 ≤ Ns ∈ N+ non-zero entries in its spreading sequence,

2 ≤ βNs ∈ N+ users shoud be assigned in the same RE;
– spreading matrix S is assumed to converge to a bipartite Galton-Watson tree
in the LSL (see ([13], Theorem 2) for a full description).
To effectively induce non integer values of Ns and βNs , one may employ time-
sharing between different (Ns, βNs ) points in the admissible set to achieve the
same total throughput as mentioned in ([13], Remark 4).
• Ns = 1: The regularity imposes β ∈ N+ users per each RE, that is equivalent
to having a set of N parallel Gaussian MAC channels [1]. This observation may
bring more insight on the behavior of the regular with respect to optimum and
linear receivers.
Figure 3 shows the achievable rates of MMSE, ZF, SUMF receivers for the dense
(Ns = N), LMMSE receiver for the regular low-dense schemes with Ns = {1, 2, 5}.
The linear receiver for the irregular low-dense NOMA with the typical case Ns = 1
is also shown for comparison. In contrast to the irregular counterpart, achievable
rates for the regular low-dense NOMA, which are superior to all other cases, grow
gradually for lower values of Ns < N, and reach the ultimate rate (Cover-Wyner
bound) when Ns = 1.
The reason that makes the optimal spectral efficiency of the irregular low-dense to
be lower than the dense case may be caused by the random nature of user-resource
allocation, leading to a condition in which some users are not assigned with any
RE, while some REs are left unused. On the other hand, the regularity feature of the
regular low-dense NOMA contributes to increasing the optimal spectral efficiency
by employing user-mapping intentionally. Nonetheless, this also imposes as a direct
consequence additional practical challenges in having some kind of coordination
while allocating the resources to users [12, 13].
For linear receivers, a remarkable observation from low-dense NOMA with Ns = 1
can be given: capacity of regular low-dense NOMA outperforms all the rest when
β 6 1 (OMA area), particularly to the typical setting when Ns = 1; while in the
overloaded regime (NOMA area), there is an intersection where capacity of irregular
low-dense NOMA with Ns = 1 outperforms all other cases. By numerical equation
solving, the exact value of the intersection is located at β = 1.232, from which
irregular low-dense NOMA with Ns = 1 dominates those of dense NOMA, as well
as with all other degrees of sparseness (Ns > 1) till about β ≈ 5, and then tend to
converge for β > 5 (with the negligible gap of about 5% at β = 5). These observed
results can be used as a driving rationale in system design.
6
irreg R lin Ns= 1
reg C LMMSE Ns = 1
5 reg C LMMSE Ns = 2
reg C LMMSE Ns = 5
dense
4 C SUMF
C dense
R (bits/s/Hz)
MMSE
C dense
3 ZF
0
0 1 2 3 4 5
β = K/N
Fig. 3: Achievable rates (bits/s/Hz) of dense NOMA (Ns = N ) with SUMF, MMSE, ZF receivers
vs. low-dense NOMA (Ns = 1, 2, 5) with linear receivers as a function of β for fixed
Eb /N0 = 10 [dB]
4 Conclusion
Motivated by the key challenge of finding and analyzing theoretical bounds for
NOMA in massive communications, this chapter sheds some light on the relationship
between achievable rates and NOMA parameters, such as load factor, degree of
sparseness and regularity. A unified framework for several code-domain NOMA
schemes was presented. The analytical framework, built on the traditional DS-CDMA
model, proved to be flexible enough for representing several code-domain dialects,
and, in particular, addressed properties of a fundamental element of the model, that
is, the representation matrix S.
Theoretical investigations were interpreted in the LSL for both optimum and linear
receivers, based on closed-form expressions existing for three distinctive cases, that
are, dense vs. regular low-dense and irregular low-dense NOMA, corresponding to
DS-CDMA (Ns = N) [38] vs. LDS-CDMA (Ns = 1) [23] and regular sparse NOMA
(2 ≤ Ns ∈ N+, βNs ∈ N+ ) [13]. For any value of load, low-dense NOMA cases
were shown to be more spectral-efficient than dense ones. For optimum receivers,
achievable rates of the regular low-dense are higher than the irregular low-dense
and dense NOMA regardless of load. To this end, the system must be constrained
to have exactly Ns REs per user and βNs ∈ N users per resource; this imposes either
central scheduling or users coordination. For linear receivers, spectral efficiency
of regular low-dense NOMA was proved to be higher than all the other cases in
the underloaded regime, while spectral efficiency of irregular low-dense dominated
other NOMA cases in the overloaded systems, particularly when the system load
β is within an interval that is about [1.2, 5]. When Ns increases, that is sparseness
decreases, achievable rates of low-dense cases rapidly converged to achievable rates
of the dense case, as soon as Ns = 2.
In conclusion, by changing the spreading strategy from dense to low-dense,
specific theoretical limits hold, showing that, to obtain higher achievable rates for
linear decoders while still enjoying the lower receiver complexity, it is advisable
to adopt sparse communications, and in particular irregular extreme low-dense
schemes when systems are overloaded and regular extreme low-dense cases in the
underloaded regime.
References
1. M. T. P. Le, G. C. Ferrante, G. Caso, L. De Nardis, and M.-G. Di Benedetto, “On information-

theoretic limits of code-domain NOMA for 5G,” IET Communications, vol. 12, no. 15, pp.
1864–1871, 2018.
2. Z. Ding, X. Lei, G. K. Karagiannidis, R. Schober, J. Yuan, and V. K. Bhargava, “A survey on
non-orthogonal multiple access for 5g networks: Research challenges and future trends,” IEEE
J. Sel. Areas Commun., vol. 35, no. 10, pp. 2181–2195, Oct. 2017.
3. 3rd Generation Partnership Project (3GPP) Portal, 2017, [Online]. Available:
https://portal.3gpp.org/, accessed on May 09, 2018.
4. S.-Y. Lien, S.-L. Shieh, Y. Huang, B. Su, Y.-L. Hsu, and H.-Y. Wei, “5G New radio: Waveform,
Frame Structure, Multiple Access, and Initial Access,” IEEE Trans. Commun., vol. 55, no. 6,
pp. 64–71, June 2017.
5. Y. Liu, Z. Qin, M. Elkashlan, Z. Ding, A. Nallanathan, and L. Hanzo, “Nonorthogonal multiple
access for 5g and beyond,” Proceedings of the IEEE, vol. 105, no. 12, pp. 2347–2381, Dec
2017.
6. L. Dai, B. Wang, Y. Yuan, S. Han, C. l. I, and Z. Wang, “Non-orthogonal multiple access
for 5G: solutions, challenges, opportunities, and future research trends,” IEEE J. Sel. Areas
Commun., vol. 53, no. 9, pp. 74–81, Sept. 2015.
7. R. E. Learned, A. S. Willsky, and D. M. Boroson, “Low complexity optimal joint detection for
oversaturated multiple access communications,” IEEE Trans. Signal Process., vol. 45, no. 1,
pp. 113–123, Jan. 1997.
8. J. G. Andrews, “Interference cancellation for cellular systems: a contemporary overview,”
IEEE Wireless Commun. Mag., vol. 12, no. 2, pp. 19–29, 2005.
9. Z. Ding, Y. Liu, J. Choi, Q. Sun, M. Elkashlan, I. Chih-Lin, and H. V. Poor, “Application
of non-orthogonal multiple access in LTE and 5G networks,” IEEE Commun. Mag., vol. 55,
no. 2, pp. 185–191, Feb. 2017.
10. S. Shamai and S. Verdú, “The impact of frequency-flat fading on the spectral efficiency of
CDMA,” IEEE Trans. Inf. Theory, vol. 47, no. 4, pp. 1302–1327, May 2001.
11. J. Raymond and D. Saad, “Sparsely spread CDMA- a statistical mechanics-based analysis,” J.
Phys. A: Math. Theor., vol. 40, no. 41, pp. 12 315–12 333, 2007.
12. O. Shental, B. M. Zaidel, and S. S. Shitz, “Low-density code-domain NOMA: Better be
regular,” in Proc. IEEE Int. Symp. Inf. Theory (ISIT), 2017, pp. 2628–2632.
13. B. M. Zaidel, O. Shental, and S. Shamai, “Sparse NOMA: A Closed-Form Characterization,”
in Proc. IEEE Int. Symp. on Inf. Theory (ISIT), Vail, CO, USA, June 2018, pp. 1106–1110.
14. J. A. F. Ross and D. P. Taylor, “Vector assignment scheme for m + n users in n-dimensional
global additive channel,” Electronics Letters, vol. 28, no. 17, pp. 1634–1636, 1992.
15. ——, “Multiuser signaling in the symbol-synchronous AWGN channel,” IEEE Trans. Inf.
Theory, vol. 41, no. 4, pp. 1174–1178, 1995.
16. Z. Yuan, G. Yu, W. Li, Y. Yuan, X. Wang, and J. Xu, “Multi-user shared access for internet of
things,” in Proc. IEEE Veh. Technol. Conf. (VTC-Spring), Nanjing, 2016, pp. 1–5.
17. S. Chen, B. Ren, Q. Gao, S. Kang, S. Sun, and K. Niu, “Pattern Division Multiple Access-
A Novel Nonorthogonal Multiple Access for Fifth-Generation Radio Networks,” IEEE Trans.
Veh. Technol., vol. 66, no. 4, pp. 3185–3196, Apr. 2017.
18. J. Zeng, B. Li, X. Su, L. Rong, and R. Xing, “Pattern division multiple access (PDMA)
for cellular future radio access,” in Proc. IEEE Int. Conf. on Wireless Commun. and Signal
Process. (WCSP), Nanjing, China, 2015, pp. 1–5.
19. 3rd Generation Partnership Project (3GPP), Non-orthogonal multiple access candidate for NR,
TSG RAN WG1 #85, Nanjing, China, May 2016, proposed by Samsung.
20. K. Li, X. Wang, and L. Ping, “Analysis and optimization of interleave-division multiple-access
communication systems,” IEEE Trans. Wireless Commun., vol. 6, no. 5, May 2007.
21. L. Ping, L. Liu, K. Wu, and W. K. Leung, “Interleave division multiple-access,” IEEE Trans.
Wireless Commun., vol. 5, no. 4, pp. 938–947, 2006.
22. R. Hoshyar, F. Wathan, and R. Tafazolli, “Novel Low-Density Signature for Synchronous
CDMA Systems over AWGN channel,” IEEE Trans. Signal Process., vol. 56, no. 4, pp. 1616–
1626, Apr. 2008.
23. G. C. Ferrante and M.-G. Di Benedetto, “Spectral efficiency of random time-hopping CDMA,”
IEEE Trans. Inf. Theory, vol. 61, no. 12, pp. 6643–6662, Dec. 2015.
24. M. T. P. Le, G. C. Ferrante, T. Q. S. Quek, and M.-G. Di Benedetto, “Fundamental Limits of
Low-Density Spreading NOMA with Fading,” IEEE Trans. Wireless Commun., 2018.
25. X. Dai, S. Chen, S. Sun, S. Kang, Y. Wang, Z. Shen, and J. Xu, “Successive interference
cancelation amenable multiple access (SAMA) for future wireless communications,” in Proc.
IEEE Int. Conf. Commun. Sys. (ICCS), Macau, 2014, pp. 222–226.
26. J. Choi, “Low density spreading for multicarrier systems,” in Spread Spectrum Techniques and
Applications, 2004 IEEE Eighth International Symposium on, 2004, pp. 575–578.
27. L. Wen, R. Razavi, M. Imran, and P. Xiao, “Design of joint sparse graph for ofdm (jsg-ofdm)
system,” IEEE Trans. Wireless Commun., vol. 14, no. 4, pp. 1823–1836, Apr. 2015.
28. H. Nikopour and H. Baligh, “Sparse code multiple access,” in Proc. IEEE Annu. Int. Symp.
Personal, Indoor, and Mobile Radio Commun. (PIMRC), London, UK, 2013, pp. 332–336.
29. M. Zhao, S. Zhou, W. Zhou, and J. Zhu, “An Improved Uplink Sparse Coded Multiple Access,”
IEEE Commun. Lett., vol. 21, no. 1, pp. 176–179, Jan. 2017.
30. 3rd Generation Partnership Project (3GPP), Low code rate and signature based multiple access
scheme for New Radio, TSG RAN1 #85, Nanjing, China, May 2016.
31. ——, Considerations on DL/UL multiple access for NR, TSG RAN WG1 #84bis, Busan,
Korea, Apr. 2016.
32. ——, Non-orthogonal multiple access for New Radio, TSG RAN WG1 #85, Nanjing, China,
May 2016, proposed by Nokia, Alcatel-Lucent Shanghai Bell.
33. ——, New uplink non-orthogonal multiple access schemes for NR, TSG RAN WG1 #86,
Gothenburg, Sweden, Aug. 2016.
34. J. Zhang, X. Wang, X. Yang, and H. Zhou, “Low density spreading signature vector extension
(lds-sve) for uplink multiple access,” in Proc. IEEE Veh. Technol. Conf. (VTC-Spring), Toronto,
Canada, 2017, pp. 1–5.
35. S. D. Sosnin, G. Xiong, D. Chatterjee, and Y. Kwak, “Non-Orthogonal Multiple Access with
Low Code Rate Spreading and Short Sequence Based Spreading,” 2017, pp. 1–5.
36. 3rd Generation Partnership Project (3GPP), Initial views and evaluation results on non-
orthogonal multiple access for NR uplink, TSG RAN WG1 #84bis, Busan, Korea, Apr. 2016.
37. ——, Candidate NR multiple access schemes, TSG RAN WG1 #84bis, Busan, Korea, Apr.
2016.
38. S. Verdú and S. Shamai, “Spectral efficiency of CDMA with random spreading,” IEEE Trans.
Inf. Theory, vol. 45, no. 2, pp. 622–640, Mar. 1999.
39. X. Dai, Z. Zhang, S. Chen, S. Sun, and B. Bai, “Pattern Division Multiple Access (PDMA): A
New Multiple Access Technology for 5G,” IEEE Wireless Commun. Mag., vol. 25, no. 2, pp.
54–60, Apr. 2018.
40. A. M. Tulino, L. Li, and S. Verdú, “Spectral efficiency of multicarrier CDMA,” IEEE Trans.
Inf. Theory, vol. 51, no. 2, pp. 479–505, Feb. 2005.
41. A. Montanari and D. N. C. Tse, “Analysis of belief propagation for non-linear problems: The
example of CDMA (or: How to prove tanaka’s formula),” in Proc. IEEE Inf. Theory Workshop
(ITW), Punta del Este, 2006, pp. 160–164.
42. M. Yoshida and T. Tanaka, “Analysis of sparsely-spread CDMA via statistical mechanics,” in
Proc. IEEE Int. Symp. Inf. Theory (ISIT), 2006, pp. 2378–2382.
43. D. Guo and C.-C. Wang, “Multiuser detection of sparsely spread CDMA,” IEEE J. Sel. Areas
Commun., vol. 26, no. 3, pp. 421–431, Apr. 2008.
5G Trials in Italy
Andrea Abrardo
Abstract In March 2017 the Italian Ministry of Economic Development (MISE)

issued a public notice to open the procedure for the acquisition of project proposals
for the realization of pre-commercial trail programs in the frequency spectrum of 3.6
- 3.8 Ghz. Since then, many 5G trial programs have been launched in Italy involving
several operators, private companies, research centers, and public institutions. In this
chapter, after an introduction that highlight the importance of 5G trials for boosting
the interest of all potential stakeholders in 5G, we describe the different actions that
followed the MISE public notice in the cities of Milan, Prato-L’Aquila and Bari-
Matera and in other Cities not directly involved in the MISE pre-commercial trial
program.
1 Introduction
1.1 5G: a new paradigm shift
The concept of a fifth generation of mobile telecommunications (5G) has attracted a

growing interest at all levels, to such an extend that everyone is being talking about
5G far long before commercial 5G services are available.
The reason for such an interest resides mainly in the high expectations generated
by 5G, not only on the telecommunications industry, but also on a vast range of
different industrial areas. As an example, a survey conducted by Ericsson [1] on 900
decision makers in large companies across 10 key industries, reveals that more than
Andrea Abrardo
Andrea Abrardo, Department of Information Engineering and Mathematical Sciences, Siena, Italy,
e-mail: abrard@dii.unisi.it
Name of Second Author
Name, Address of Institute e-mail: name@email.address
291
292 Andrea Abrardo
three-quarters of respondents predicted that their sector would take advantage of 5G.
The 10 industries covered in the study include energy and utilities, manufacturing,
public safety, healthcare, media and entertainment, public transport, automotive,
financial services, retail, and agriculture.
As a matter of fact, the 5G development process established a change of paradigm
with respect to previous generations of mobile communications standards. In par-
ticular, at an early stage of 5G development, the discussion on new technologies
has been preceded by a thorough investigation on the use cases and needs that 5G
should fulfill. Then, after a general agreement about the use cases, the 3G Partner-
ship Project (3GPP) international standardization body [2] discussed and evaluated
candidate technologies to satisfy the required bit rates, latencies and concentration
of nodes. This process is still ongoing, and the initial 5G standardization framework
called 5G New Radio (NR) [3] has been frozen in 2018.
1.2 The importance of experimentation
Despite commercial 5G networks will not roll out before 2019, there’s a lot of testing
activities being conducted all around the World. Putting it in general terms, every
new technology is first tested in the labs, through early trial equipment. However,
despite extensive simulations carried out in the labs may assess the validity of a given
technology, it is often unclear how a new technology behaves in real environments.
On the other hand, the forthcoming 5G mobile networks are expected to bear
a multitude of new technologies such as (i) the use of mmWaves, (ii) massive
Fig. 1 The 5G scenario

5G Trials in Italy 293
use of SDN and NFV to make the network more flexible and adaptive, (iii) the
introduction of computational elements close to the users (in the edge) regulated
by network slicing techniques to support specialized services, (iv) the adoption of
multiple and heterogeneous connectivity techniques, including (v) unconventional
connectivity schemes, e.g., the use of D2D communications, and (vi) the support for
IoT architectures.
Moreover, since one of the main targets of 5G is the possibility of enabling
vertical sectors such as e-Health, factories of the future, energy and automotive, the
trial activity must also be intended to prove the attainability of such an ambitious
goal through the implementation of a variety of use cases covering different sectors.
Eventually, one of the ultimate purposes of 5G trials is that of boosting the interest
of industry and public administration in 5G as an enabler of new business and new
services to citizens, as well.
To sum up, the 5G trial activities require a continuous experimental validation,
which is a hard achievement considering the heterogeneity of the 5G scenarios.
As a matter of fact, the validation of the ambitious network-oriented 5G targets
in real operational systems and the impact on end-user performance remains an
open issue. For this reason, the experiences matured during experimental results,
and the correspondent dissemination of the results, both in terms of measurement
methodologies and testbed deployments is a fundamental enabler for the development
and deployment of 5G networks.
2 The Italian scenario
On the 14th of September 2016 the European Commission called on Member States
to identify by 2018 at least a city where start testing of 5G (Action Plan for the 5G).
Hence, the 5G trial process started in Italy at the beginning of 2017 with the goal
of going further the EU guidelines: the 16th of March 2017 the Italian Ministry of
Economic Development (MISE) [4] published a public notice to open the procedure
for the acquisition of project proposals for the realization of pre-commercial trail
programs in the frequency spectrum of 3.6 - 3.8 Ghz. The projects should have been
completed in four years in the following geographical areas:
• Milan
Fig. 2 The five cities involved in the MISE 5G testing activities

294 Andrea Abrardo
• Prato-L’Aquila
• Bari-Matera
From that point on, the journey to provide Italy with the technology for fifth genera-
tion mobile networks, begins. The cities of Milan, Prato and Bari have been selected
through the following criteria: geographic distribution, capillary action of ultra-fast
connectivity, availability of frequencies in the band from 3.7 to 3.8 GHz, proximity
the European corridors. In addition to these cities, L’Aquila and Matera were also
included: the first because of its post-earthquake reconstruction phase, the second
thanks to the prevision (that came true) to be the next European Capital of Culture
in 2019. After the publication of the public notice, several applications have been
received by the MISE and three project proposals were elected to start the 5G testing
activities in 2018 in the three areas:
• Metropolitan area of Milan, assigned to Vodafone Italia S.p.A.
• Prato-L’Aquila, assigned to Wind Tre S.p.A. and Open Fiber S.p.A.
• Bari-Matera, assigned to Telecom Italia S.p.A., Fastweb and Huawei Technologies
Italia S.r.L.
At the same time, the following 5G testing activities have been launched in Italy
involving different geographical areas.
March 2017: TIM announces the realization of the first 5G urban network in
Torino.
June 2017: The 5GCity project, involving the cities of Lucca, Barcelona and
Bristol, is granted with 8 million of euros within the H2020 framework. In this
context, Lucca is selected as ideal scenario for representing the nearly 1400 small
cities in Europe counting 40000-200000 residents.
September 2017: Roma Capitale and Fastweb sign an agreement for the launch
of 5G testing in Rome involving Ericsson and ZTE.
March 2018: The municipality of Genova, the Liguria Region, Liguria DIgitale,
Ericsson and TIM sign an agreement for the creation of the Digital Lab 5G at Great
Campus Erxelli specifically aimed at the realization of the new mobile wireless
network in Genova for fostering new Internet of Things scenarios.
Finally, in 2016 TIM and Ericsson launched the project 5G for Italy having the
Port of Livorno as an industrial testbed for the implementation of a General Cargo
Management System using 5G technologies.
References
1. Ericsson: The industry impact of 5G.

https://www.ericsson.com/en/networks/trending/insights-and-reports/industry-business-
impact-of-5g
2. 3GPP: The mobile broadband standard.
http://www.3gpp.org
3. 3GPP: First 5G New Radio Specifications.
http://www.3gpp.org/news-events/3gpp-news/1929-nsa_nr_5g
5G Trials in Italy 295
4. Minitero dello sviluppo economico (MISE): Progetti sperimentali 5G.

https://www.sviluppoeconomico.gov.it/index.php/it/213-normativa/notifiche-e-
avvisi/2036226-5g-avviso-pubblico-per-progetti-sperimentali
Part III
5G research in Italy
5G and Relevant Enabling Technologies
Paola Iovanna, Luca Stroppolo, Giulio Bottari
1 Introduction
The mobile industry has passed many big milestones over the last twenty years,
which have led to fundamental changes to our behaviors and society. When mo-
bile data traffic surpassed mobile voice traffic in 2009, it was difficult to know
what today’s use of mobile technology would look like. However, all the previous
mobile technology generations were basically developed to address only con-
sumer predominantly: for voice and SMS in 2G, for basic internet browsing in
3G, and higher speed data and video in 4G).
5G and the Internet of Things (IoT) promise new capabilities and use cases,
which are set to impact not only consumer services but also many industries em-
barking on their digital transformations. In fact, 5G will serve consumers and mul-
tiple industries and open new business models across industries. 2018 will be the
Paola Iovanna
Ericsson
Luca Stroppolo
Ericsson
Giulio Bottari
Ericsson
year for the first rollouts of 5G networks as well as large-scale deployments of
cellular IoT.
This change will require the combined effort of industry players, operators, and
regulators to align on technology, new services and use cases, spectrum, and
standards.
2 5G Use Cases
5G wireless access is being devel-

oped with three broad use case fam-
ilies in mind [1]: enhanced mobile
broadband (eMBB), massive ma-
chine-type communications
(mMTC) and ultra-reliable low-la-
tency communications (URLLC).
5G will kick off with eMBB as
its first use case. With the continued growth in mobile traffic, there is a need for
a more efficient technology, higher data rates, and spectrum utilization. In terms
of consumer digitalization and richer user experience, new applications based on
virtual and augmented reality will require higher bandwidths and lower latency.
Key elements of eMBB are: i) massive mobile connectivity as demand for mo-
bile broadband will continue to increase; ii) high data rates, connection density
and mobility; iii) Human-centric use cases for access to multi-media content,
e.g. 4k streaming on mobile or on-site live experiences.
mMTC means multiple sensors pinging low volume data from machines,
wearables and infrastructure. Examples are logistics trackers, temperature control,
Narrow Band IoT (NB-IoT) to enable sensors with ten times longer battery life.
Key elements: i) usage scenarios which need connectivity for millions of de-
vices; ii) each device typically transmitting a relatively low volume of non-delay-
sensitive data (low BW & not-latency critical); iii) Devices are required to be low
cost and have a very long battery life.
cMTC means supporting remote control of robots and big excavators in
mines, traffic and safety control in increasingly mobile factories with automated
5G and Relevant Enabling 301
Technologies
guided vehicles for material handling and drones for infrastructure support, re-
mote medical surgery, distribution automation in a smart grid, transportation
safety, etc. It requires real time data and actions and ultra-low latency (at milli-
second level). Key elements: i) usage scenarios which need ultra-reliable resili-
ent, instantaneous connectivity; ii) stringent requirements for capabilities such
as latency, availability, and reliability.
3 5G Spectrum and Enabling Technologies
To meet the complex and sometimes contradictory requirements of these diverse

use cases, 5G will encompass both an evolution of today’s 4G (LTE) networks
and the addition of a new, globally standardized radio access technology known
as New Radio (NR). 5G NR [2]-[3] will operate in the frequency range from
below 1 GHz to 100 GHz with different deployments. There will typically be more
coverage per base station (macro sites) at lower carrier frequencies, and a limited
coverage area per base station (micro and pico sites) at higher carrier frequencies.
Spectrum. Spectrum is fundamental for wireless communication and there is
a never-ending quest for more spectrum to meet the demands of increased capac-
ity and higher data rates. This is one of the major reasons why NR needs to exploit
also frequencies in the mm-wave range, as well as aggregation of multiple wide-
band carriers. However, the amount of licensed spectrum an operator has access
to may not be sufficient and there is typically a cost associated with obtaining a
spectrum license. Despite the less controlled interference situation, unlicensed
spectrum is of interest, primarily as a complement to licensed spectrum but also
on its own in local deployments. So, to provide high service quality and optimal
reliability, licensed spectrum will continue to be the backbone of the wireless net-
work in 5G, and transmission in unlicensed spectrum will be used as a comple-
ment to provide even higher data rates and boost capacity.
The spectrum situation for 5G is far from similar around the world. Large
differences exist between countries when it comes to spectrum allocations and
availability dates. This makes every operator’s starting position for 5G unique and
tailored solutions are needed to fit all the needs. In general terms, we can expect
that initial 5G deployments will be non-standalone in low, mid and high bands,
interworking with 4G. 3.5GHz is one interesting option that is being considered
widely, while also high-band millimeter wave implementations can be expected.
As traffic increases and markets mature, we can expect to see more and more
combinations of bands. Existing spectrum, in which 4G is currently deployed, will
be gradually migrated to 5G, when the increased penetration of 5G devices is high
enough to offload the load on 4G spectrum. Functionalities for both spectrum mi-
gration and combinations (both bands and technologies) are crucial for operators
to be able to make a smooth 5G evolution plan.
5G Spectrum and deployments
Multi-antenna transmission, including MIMO and beamforming, is a key part

of NR and will evolve further in release 16 of the standard Joint transmission from
multiple, geographically separated antenna panels can bring additional capacity
and improve data rates. Multi-antenna transmission already plays an important
role in current generations of mobile communication and will be even more cen-
tral in the 5G era. Massive MIMO (multiple-input and multiple-output) uses a
combination of advanced antennas with many steerable ports to significantly in-
crease the number of transmission points, spectral efficiency, and network capac-
ity and delivers faster data throughput.
Beamforming utilizes the Massive MIMO advanced antenna array to increase
the capacity and effectiveness of radio transmissions. Beamforming achieves this
Technologies
by shaping the radio signals into highly focused, steerable beams which deliver a
stronger radio signal a greater distance and with less energy. As a result, Beam-
forming and Beam steering enable higher data throughput throughout the cell, and
higher data rates at the cell edge. With Beamforming, the radio data transmission
is sent directly to the device instead of broadcasting across the entire cell which
also reduces intercell interference further improving system performance. The
number and shape of the beams is controlled dynamically to meet the application
needs.
Enhanced feedback from the terminals about the radio conditions can allow
more extensive usage of advanced multi-antenna schemes and bring gains.
Dual connectivity enhancements, for example support of asynchronous dual
connectivity, is another example of an enhancement relevant for geographically
separated cells.
With the need for higher data rates and higher capacity, the networks are likely
to become more and more dense, stressing the need for efficient backhaul from
all these base stations. Integrated access and backhaul (IAB) refers to the work
in 3GPP on using NR not only for the access link but also for the backhaul link
connecting the base stations to the core network. For example, smaller base sta-
tions may be deployed and connect wirelessly using NR to macro base stations
with fixed backhaul.
Fronthaul for 5G: eCPRI. To meet the needs for 5G, the new industry stand-
ard 5G fronthaul—eCPRI—will improve bandwidth efficiency, increase capaci-
ties and lower latencies. Ericsson has implemented the market’s first commercial
eCPRI interface in Ericsson Radio System. Its bandwidth scales flexibly with user
traffic and allows for a 10x reduction of bandwidth on the interface. eCPRI will
use standard Ethernet connections over dedicated fiber. By using the eCPRI inter-
face, Ericsson has moved beamforming processing from the baseband to the radio.
This allows for a much simpler Massive MIMO deployment and provides the
flexibility needed in real-life site environments.
Ultra-lean radio-access design is important to achieve high efficiency in 5G
networks. The basic principle of ultra-lean design can be expressed as: minimize
any transmissions not directly related to the delivery of user data. Such transmis-
sions include signals for synchronization, network acquisition and channel esti-
mation, as well as the broadcast of different types of system and control infor-
mation. Ultra-lean design is especially important for dense deployments with
many network nodes and highly variable traffic conditions. However, lean trans-
mission is beneficial for all kinds of deployments, including macro deployments.
Virtualized RAN for 5G. Some of the RAN functions that are less sensitive
to time delay which are hosted within the baseband units can become virtualized
to increase network flexibility., For example, the multipath-handling function that
is the anchor point for dual connectivity in 5G can be further optimized. By mov-
ing this function higher up in the network, “tromboning” traffic delays and inef-
ficiencies can be avoided.
The 5G standardization has been accelerated with first 5G NR standard fi-
nalized in Dec 2017 and completed in June 2018. First commercial 5G networks
and devices based on the 3GPP standards are expected in 2018. The first very few
5G devices will likely be introduced towards the end of 2018. Ericsson estimates
the number of subscriptions reaching one billion by the end of 2023.
In early 2018 Ericsson has launched the industry’s first 5G NR-capable ra-
dio, called Ericsson AIR 6468. It features 64 transmit and 64 receive antennas
enabling it to support our 5G plug-ins for both Massive MIMO and Multi-User
MIMO. The high-performance beamforming, required for Massive MIMO, is en-
abled through the use of a split Cloud RAN architecture, which brings the required
intelligence and scalability to this new radio. And, the AIR 6468 is designed for
compatibility with the 5G NR standard while also supporting LTE.
In September 2018 Ericsson has announced the first 3GPP-compliant 5G
New Radio (5G NR) millimetre-wave (mmWave) over-the-air call using a
smartphone form-factor mobile device in partnership with Qualcomm. The call
was made in Ericsson's Kista, Sweden, lab, utilising the 39GHz mmWave spec-
trum band and Ericsson's commercial 5G NR Air 5331 radio and baseband prod-
ucts, as well as a mobile test device packing a Qualcomm Snapdragon X50 5G
modem and RF subsystem.
4 Ericsson 5G Application Scenarios
Ericsson has identified six industrial social and industrial applications [4] in which
5G will bring substantial revolutions. The following sections illustrate such
Technologies
applications detailing the beneficial impact of 5G and the relevant main technol-
ogy enablers.
4.1 Broadband and Media Everywhere
This scenario shows how it will be possible to communicate in crowded or remote

areas at high speed, thanks to a decrease in latency and the increase in data rates.
Thanks to 5G, every mobile user will experience high-quality broadband service
both upload and download.
With 5G, users will experience broadband access in crowded areas like con-
certs, sporting events and festivals, alleviating issues with capacity, interference
and reliability. 5G customers will also enjoy 4K movies downloaded in just sec-
onds without a Wi-Fi connection. While live TV broadcasts and sporting events
will become immersive viewing experiences, as if you were at the event in real-
life. For customers, 5G will also maximize their experience in both indoor and
outdoor connectivity and offer high QoS broadband even in challenging network
conditions.
Technology
enablers
5G radio access • Capacity and full mobility
• Improved beam forming
• Massive MIMO
• Broadcast
• Carrier aggregation
• New high frequency spectrum
5G core network • QoS support for e.g. emergency/safety related communica-
tion
• Roundtrip latency in 1 ms range
• Cloud based flexible deployment of media services
• Network slices all optimized for mobile broadband and media
delivery
5G management & • Congestion handling per subscriber/service or based on usage
orchestration
• Dynamic allocation of resources according to traffic variation
• Reduce load on transport links and central processing units
4.2 - Smart vehicles and transport
Sensors embedded in roads, railways, airfields and vehicles will communicate

with each other through the 5G network. Ericsson is fully aware that this use case
is focused on massive machine type communication and is working to ensure that
the 5G network has the necessary high coverage and low power consumption.
When transportation and vehicles are equipped with 5G connectivity, it will
revolutionize the way we travel. Vehicle-to-vehicle and vehicle-to-infrastructure
communication will make roads safer and more environmentally friendly, while
allowing buses and public transportation to run more efficiently. New services
and business models can be supported considering sensors embedded in roads,
railways and airfields to communicate to each other and/or with smart vehicles.
Technology
enablers
5G radio access • Massive density
• Device cost Mobility & latency
• Significantly reduced signaling overhead – device energy con-
sumption
• Soft-SIM or no-SIM operation for sensor type devices
5G core network • Integrate public infrastructure network within network slices

• Support for pub/sub message-oriented communication
Technologies
5G management & • Orchestration of a big amount of data and input interfaces

orchestration
• Common view for all the utility/infrastructure suppliers
• Different user profiles to access the same network
4.3 - Critical services and infrastructure control
5G brings high reliability and low latency required to control critical services and
infrastructure. This unlocks new opportunities for public safety, government, city
management and utility companies.
5G will be a key enabler of the future digital world. It will bring about new
service capabilities for industrial stakeholders thanks to the unprecedented on-
demand performance and real-time reactivity. For example, energy and water util-
ities will can connect to millions of networked devices, taking real-time, intelli-
gent and autonomous decisions.
Technology enablers
5G radio access • Massive density
• Device cost
• Significantly reduced signaling - device energy con-
sumption
• Soft-SIM or no-SIM operation for (at least) sensor
type devices.
• Always available and basic services e.g. emergency
calls
5G core network • Integrate infrastructure network within network
slices
• Reliability
5G management & orche- • Orchestration of a big amount of data and input in-
stration terfaces
• Common view for all the utility/infrastructure sup-
pliers
• Co-relate and process different inputs for fast re-
sponse
4.4 - Critical control of remote devices
5G introduces the ability to remotely control devices and heavy machinery, thus
unlocking the potential for improved medical services and surgeries that would
otherwise unavailable to rural communities. 5G would also allow for heavy ma-
chinery to be controlled from a distance, making hazardous situations safer for
humans.
Amongst other benefits, 5G will allow for heavy machinery to be controlled
remotely. This lowers the risk of injury in hazardous environments and means that
work can be completed in a more effective way. Similarly, industries like manu-
facturing and mining will experience better efficiency and reduced costs.
Technologies
5G radio ac- • High node/service availability at least 99.999% node availability

cess
• Uplink capacity for high quality video
• Mobility & latency
• Best use of licensed and unlicensed technologies
5G core net- • QoS functions and network slices

work
• 99.9% accessibility and retain ability
5G manage- • Improve response time for diagnostic
ment &
• Meet real-time constraints
orchestration
• Estimate and report about the achieved reliability of a connection
(per user, per service)
4.5 – Human-Machine Interaction
The high performance of 5G networks will make IoT more accessible by humans,
to enhance the awareness of the context in which people live. Ericsson 5G system
allows for the context awareness that allows you to fill the gap between people
and IoT.
5G will become the backbone of IoT, connecting devices in ways we never
thought possible. Users will experience smart cars that are capable of communi-
cating with traffic lights; augmented reality and 360 degree immersive gaming
and movies; and transmitting touch and texture to realize the tactile internet.
The IoT applications that 5G will help enable is truly limitless.
5G radio ac- • High node/service availability at least 99.999% node availability
cess
• Uplink capacity for high quality video
• Extremely low latency
• Best use of licensed and unlicensed technologies
5G core net- • QoS functions and network slices 99.9% accessibility and retain
work ability for comm. services
5G manage- • Improve response time for diagnostic questions
ment &
• Meet real-time constraints
orchestration
• Estimate and report about the achieved reliability of a connection
(per user, per service)
4.6 – Sensor Networks
5G technology will expand business opportunities through monitoring, tracking

and automation capabilities on a large scale - from connected farms and agricul-
ture to smart cities and buildings.
5G will allow the IoT (Internet of Things) era to become a reality by breaking
constraints and capturing information in new ways. For instance, sensors will be
able to be implemented throughout farms allowing for crops to communicate
moisture and fertilization needs. While utility companies will be able to monitor
and report on energy useage. Thus, improving the reliability and sustainability of
energy production and distribution.
Technologies
5G radio access • Device cost

• Significantly reduced signalling overhead – device en-
ergy consumption
• Soft-SIM or no-SIM operation
• Reliability
• Coverage & spectrum efficiency
5G core network • QoS functions
• Network slices
5G management & orche- • Orchestration of a big amount of data and input inter-
stration faces
• Common view for all the utility/infrastructure suppliers
• Co-relate and process different inputs for fast re-
sponse
• Data analytics – business inteligence
5 The 5G for Italy program
TIM and Ericsson announced in June 2016 [5] the launch of the "5G for Italy"
program aimed at creating an open ecosystem for research and implementation of
innovative projects enabled by 5G networks - the next generation of mobile tech-
nology - to accelerate digitization of the country.
The initiative brings together industries, institutions, universities, research
centers, local administrations and small and medium-sized businesses to develop
and test new services and pilot projects using 5G technology. The main areas of
analysis are the Smart City, the Internet of Things, Industry 4.0, Transport and
Smart Agriculture.
The "5G for Italy" program is an opportunity to develop applications that can
be used by both citizens and industries in the Networked Society.
In a broader scenario, the collaboration of the companies participating in the
project envisage monitoring and identifying emerging trends in the Italian and
international scenario and analyzing and studying the new technological and busi-
ness opportunities that 5G technology will introduce in the market starting from
2020.
As global standardization will be a crucial element for the success and rapid
deployment of the new 5G technology, it will also be the program's objective to
increase the companies' competences on the evolution of current mobile networks
towards 5G, to address and support the development of a unique "5G" standard.
Comau and the factory of the future

TIM, Ericsson and Comau partnership is aimed to build the factory of the future
within the Industry 4.0 paradigm with specific applications in the areas of Cloud
Robotics for Smart Manufacturing. This specific use case is detailed in the next
section.
The Port of the future in Livorno
Technologies
One of the first projects in the 5G for Italy program is based on the collaboration
between TIM, Ericsson, CNIT (National Inter-University Consortium for Tele-
communications) and the Port of Livorno. The main scope is to implement the
first ‘Connected Port’ where Augmented and Virtual reality technology, sup-
ported by the 5G mobile infrastructure will be used to optimize the transportation
logistics with the aim to improve the port efficiency and throughput.
Italdesign and Smart Transportation
TIM, Ericsson e Italdesign are working to extend 5G capabilities to V2x applica-
tions such as Assisted Driving support with holographic call, Autonomous Driv-
ing, Infotainment.
Seikey: drones for Public Safety
TIM, Ericsson and Seikey are collaborating on 5G network support for drone re-
mote flight control. Main use cases are related to drone owner tracking, irregular
flight detection, safety in public parks and business delivery operations.
6 5G for Industry 4.0: the Comau Case
Since 2016, Ericsson has partnered with Comau, a world leader in industrial au-
tomation to experiment the factory of the future paradigm in a real industrial con-
text. More specifically, Partners have explored low-latency network and edge
cloud for manufacturing plants. Details about these experiments, including the
results, were published in Ericsson Technology Review [6].
The joint research with Comau addresses the evolution towards Industry 4.0.
The critical requirements of wireless networking in a plant have been the main
challenge, as industrial protocols demand tight radio performances with high re-
liability, availability, and security. Wireless connected sensors (pressure, temper-
ature, vibrations, cameras) feed expert systems to determine what is happening,
to predict future issues, and to prevent failures. Adaptability is addressed by ena-
bling and implementing the centralization and virtualization of controllers, at all
levels: plant, production lines, working robotic cells, single robots, actuators.
Having a central, global control entity facilitates a fast reaction and an autono-
mous response to any event and, when needed, to invoke the action and decision
of the human supervisor.
The trend for “personal” product customization is growing, along with a pref-
erence for online purchasing. Therefore, current processes need to be adapted to
be more flexible and customizable, while still protecting initial investments into
the production line. High speed wireless infrastructure, such as 5G networks, can
help modification, required by customized products, of an original equipment
manufacturer (OEM) machine with minimal impact.
The digitization of factory operations enabled by IoT technologies promises
to make that happen. Digital tools will be able to monitor and control all tools of
production, collecting data from thousands of sensors to create a digital image of
the product being realized, usually referred to as a “digital shadow.” Once a digital
shadow has been created for a physical product and bears its specific DNA, it is
possible to manufacture that product more efficiently and with higher quality in
the digitized production facility. In this way, it is possible to optimize the manu-
facturing process, detect quality issues early to prevent defects at the end of the
production line and make continuous improvements. It is also possible to carry
out predictive and preventive maintenance.
A mature plant has up to 1000 moving parts plus many more devices and
sensor. Data traffic is no longer supportable by existing wired internal networks,
largely based on legacy Ethernet networks. Current plant networks start to have
bottlenecks.
There are three alternative options to solve the problem:
1. Add a second wired network in support of the legacy network: possible but
not easy to integrate both the networks together
2. Complete re-cabling of the plant, for example using optical fibers
3. Going wireless, using a 4G/5G cellular technology in licensed spectrum
The third option is preferred by Comau because:
▪ It reduces the cost for connection of various elements of the line, reducing the
commissioning cost for plant construction/upgrade.
▪ It facilitates using a plethora of new wireless sensors, easy to attach on robots
even in legacy plants: sensors with long lasting batteries and communicating
few data in real time. Example of sensors are: pressure, temperature, vibration
(IMU). Correlating data from these sensors would enhance monitoring and
prevention.
▪ It facilitates connecting new robots and moving objects without complex and
inflexible cabling. Cabling are easily subject to attrition especially in fast
moving robots.
▪ It enables connecting AGVs that will replace conveyor belts in future.
Technologies
▪ It is more scalable with the number of “objects” that are and will be connected
in a plant.
Wi-Fi is not viable in a factory environment for various motivations:
▪ It is a “best effort” communication technology on a shared spectrum, not ro-
bust to the co-channel interference due to the presence of multiple access
points competing for bandwidth. The phenomenon is well controlled in cellu-
lar technologies operating on a licensed spectrum.
▪ The scalability required in a factory (e.g. for adding new sensors and devices)
cannot be ensured by Wi-Fi because performances degrade with the number
of connected objects. In fact, too many clients, converging on a single access
point, bring to a detrimental contention loss.
▪ It’s not easy to make a stable Wi-Fi coverage planning of the plant because a
“signal strength mapping” is quite inaccurate and can be dynamically affected
by “objects” moving around.
Communications among all the factory elements must work in a challenging en-
vironment characterized by electromagnetic interferences and distributed over
a large area that could span several buildings. While LTE connectivity is robust
and capable enough to cope with that environment today, stringent latency re-
quirements will soon demand 5G connectivity. In particular, 5G is the target tech-
nology truly capable of delivering the performance needed to complete moving
and virtualizing robot controller functionalities in cloud, with the required security
and stability.
5G, with dedicated cellular radio coverage, will ensure:
▪ an almost immediate response from the network (= very low latency, up to 1
millisecond)
▪ slicing to serve different needs in the plant with same radio installation: e.g.
critical machine type communication for robot control, NB-IoT for capturing
data from sensors, broadband for HD cameras used to detect parts alignment
before robot pick up and AGVs…
▪ maximum security and stability in wireless data transmission
▪ a standard communication technology
The combination of wireless sensors and high-capacity communication networks
such as 4G and 5G plays a key role in this context, by enabling data collection
from shop-floor level (production lines) and data transfer to cloud systems for
continuous monitoring and control.
The following scheme illustrates the main components of the proof-of-con-
cept. Both radio core and cloud servers are located on premises. A VPN connect
the demo area with the TIM central office to enable the use of the licensed spec-
trum (dedicated to the experiment).
High level sketch of the Comau setup, including link to operator central office
Virtual controllers that combine control, data logging and alarms into a
cloud platform also help the process of digitization and save cost, panel space and
maintenance activities compared with traditional control systems. They can con-
trol a wide variety of production tools and are also a solution for remotely located
machines and portable systems that can run standalone.
References
1 ITU-R M.2083-0, “IMT Vision – Framework and overall objectives of the future devel-
opment of IMT for 2020 and beyond”, September 2015.
2 Stefan Parkvall “5G NR release 16 – start of the 5G Evolution”, Ericsson Research
Blog, https://www.ericsson.com/research-blog/5g-nr-release-16-start-5g-evolution/
3 Ali A. Zaidi, Robert Baldemair, Mattias Andersson, Sebastian Faxér, Vicent Molés-
Cases, Zhao Wang, “Designing for the future: the 5G NR physical layer”, Ericsson
Technology Review
Technologies
4 Ericsson 5G Use Cases, https://www.ericsson.com/en/5g/use-cases

5 5G for Italy press release, https://www.ericsson.com/en/news/2016/6/tim-and-erics-
son-launch-5g-for-italy-program-to-accelerate-the-countrys-digitalization
6 R. Sabella, A. Thuelig, M. C. Carrozza , Massimo Ippolito,“Industrial automation ena-
bled by robotics, machine intelligence and 5G”, Ericsson Technology Review
5G-MoNArch
Mauro Renato Boldi and the Project Partners2
1 Description and objectives of the project

The overall concept of 5G-MoNArch and its implementation are illustrated in
. The ultimate goal is the full design, implementation and evaluation of the 5G
architecture. The concepts are complemented with five key innovations required
to achieve the 5G-MoNArch objectives; these include three enabling innovations
contributing to the baseline architecture, and two functional innovations which
correspond to specific network slices.
Mauro Renato Boldi

TIM
320 Mauro Renato Boldi and the Project Partners
5G-MoNArch
Enabling innovations Common
architecture
Architecture Secure and resilient Deployment
Sea port
instantiation network slice
Cloud-enabled building testbed
Enabling common network functionality
5G-MoNArch overall architecture

protocol stack blocks
Slice specific
functions
Inter-slice
control & mgmt.
Secure and resilient
Experiment-
network functions
driven
optimization 5G-MoNArch
Functional innovations
Resource-elastic
virtual functions
Common
Previous projects architecture
building Slice specific
blocks functions
5GPPP Phase 1
architecture Architecture Resource-elastic Touristic city
instantiation Deployment
network slice testbed
Figure 1-1: 5G-MoNArch concept.
As shown in Figure 1-1, combining 5G-PPP Phase 1 architecture concepts with

the three enabling innovations results in the 5G-MoNArch “enabling common
network functionality”. This enabling functionality includes the common archi-
tectural functions needed to dynamically instantiate network slices that meet
specific use case requirements. To deploy slices that provide specialised func-
tionality, the common network functionality are complemented with two func-
tional innovations: (i) secure and resilient functions, and (ii) resource-elastic
functions. The combination of the enabling common functions with the func-
tional innovations forms the 5G-MoNArch overall architecture. This architecture
is designed to enable the integration of external novel functions (in addition to
our functional innovations), such as, e.g., that provided by other 5G-PPP pro-
jects) that may target specific requirements different from the ones in 5G-MoN-
Arch, providing thus extensibility and future-proofness.
Building on the above two functional innovations, 5G-MoNArch develops
specific NFs for two representative use cases requiring security and resilience
on the one hand, and network elasticity on the other hand. These NFs is de-
ployed in two dedicated network slices, one for each use case. These network
slices are implemented in two testbeds: the first one is a sea port testbed, which
focuses on security and resilience, and the second one is a touristic city testbed
in an urban downtown environment, which applies the resource elasticity con-
cepts.
5G-MoNarch 321
In the context of the project concept outlined above, the individual objec-
tives of 5G-MoNArch are detailed next.
Objective 1: Detailed specification and extension of the 5G architecture.

5G-MoNArch fully specifies the 5G architecture complemented with key con-
cepts and innovations derived within the project. Indeed, while 5G-PPP Phase 1
projects have developed the key concepts of the 5G architecture, they have not
provided a full architecture specification. The specification of 5G-MoNArch in-
cludes the functional control and user plane architecture as well as the manage-
ment and orchestration plane. Furthermore, attention is put to the specification
of all aspects of network slices, including their implementation in access and
core, their management, orchestration, and control. The architecture specifica-
tion includes the three enabling and two functional innovations of 5G-MoNArch,
which are detailed in the following objectives.
Objective 2: Extend the architecture design with three key enabling innovations
that enable the operation of sliced 5G networks and specific functional extensions.
To provide the desired functionality, the architecture concepts developed by 5G-
PPP Phase 1 projects need to be complemented by three architecture enablers
that are in the focus of 5G-MoNArch:
• Inter-slice control and cross-domain management, which enables the in-
terworking of different network domains operated and used by different ten-
ants while providing cross-domain service guarantees.
• Experiment-driven modelling and optimisation of a virtualised environ-
ment, which builds on specific processing models of the deployed communi-
cation platforms. This innovation leverages the implementation of 5G-MoN-
Arch architecture to gain experimental insights into the architecture and use
these insights to design optimised algorithms and functions.
• Native cloud-enabled protocol stack, which reduces dependencies within
the NFs in the protocol stack in order to enable a more flexible placement of
such functions within the network.
• The above three architecture enablers, as well as two functional innovations
described in the next objective, complement the 5G-PPP Phase 1 architecture
concepts (more details are provided in Section 1.3).
Objective 3: Develop novel functional innovations required for two key technolo-
gies required for the use cases identified.
In addition to the general enabling architecture targeted by the two objectives

above, more specific functions may be required when deploying network slices
with particular requirements. In this context, two main technology gaps have been
identified to address the performance requirements pinpointed in 5G-PPP Phase
1 and not addressed by ongoing SDO initiatives or previous projects: (i) “resili-
ence and security”, and (ii) “resource elasticity.” The novel functional innovations
required by both technologies is in the focus of 5G-MoNArch:
• The key technology resilience and security addresses the following two is-
sues: (i) the need to provide a failsafe mobile network operation which can be
maintained even under difficult circumstances such as outage of parts of the
infrastructure (long-term outage) or due to changes of the radio propagation
environment (short-term outage); and (ii) the need for guaranteed security and
privacy even in the case of shared infrastructure and resources, including non-
exclusively used spectrum, e.g., provided by public MNOs. For the latter, we
need to take into account that IoT devices may be subject to stringent power
efficiency requirements, and thus efficient lightweight security mechanisms
may be required.
• Resource elasticity addresses the need for managing information processing
and communication resources in a flexible and efficient way. For instance, in
the case of a typical urban downtown scenario, the required services may
range from augmented reality to video chats and instant messaging, each im-
posing different requirements over a certain period of time at a specific loca-
tion. To address this problem in a cost-efficient manner, (i) the mobile net-
work has to be able to assign, scale and cluster resources to those parts of the
networks where they are needed; and (ii) the NFs need to be elastic so as to
adapt to the available resources without impacting performance significantly.
Objective 4: Deployment and experimental implementation of the architecture in

two use cases.
5G-MoNArch experimentally validates the introduced key concepts and innova-
tions. The experimental validation is based on prototypes provided by the manu-
facturers of the project consortium, deployed in testbeds managed by the operators
along with vertical industries. The implementation of the 5G architecture is used
to provide “closed-loop” feedback for the specification and refinement of the 5G
architecture. To achieve this objective, the following two use cases are deployed
by 5G-MoNArch (see Section 1.3.3 for a description of the detailed testbed set-
ups):
5G-MoNarch 323
• Sea port use case: the first testbed is deployed at the Hamburg sea port to
showcase typical applications in an industrial enterprise environment. For this
use case, the “resilience and security” functional innovation described in the
previous objective is of upmost importance.
• Touristic city use case: the second testbed is deployed in the city of Venice
to showcase typical public 5G services, with a special focus on the high de-
manding applications resulting from a touristic environment (such as haptic
Internet, virtual or augmented reality). For this use case, the “resource elas-
ticity” functional innovation is of primary interest.
The above two use cases are representative of a much wider variety of sce-
narios that cover most of the performance requirements, features and technologies
discussed in 5G-PPP Phase 1 [5GAWP]: (i) the sea port use case is representative
for a private deployment where a customer requires a solution within its own
premises that provides guaranteed and secure performance and solves a specific
need; and (ii) the touristic city use case is representative for public deployments
where customers use very high resource demanding applications but may accept
that occasionally quality degrades slightly.
Objective 5: Evaluation, validation, and verification of the architecture perfor-
mance
5G-MoNArch defines a methodology and framework to evaluate, validate, and
verify the architecture based on the two testbeds deployed by the project. Building
on this framework, 5G-MoNArch quantitatively evaluates the performance of the
proposed innovations, and it validates and verify the overall architecture based on
functional, operational, and performance KPIs. This framework considers techno-
economic aspects as well, to verify the economic advantages of the proposed ar-
chitecture.
2 Italian contribution to the project
TIM is the only Italian partner in 5G-MoNArch. The research areas of TIM (be-
longing to the Wireless Access Department) involved in 5G-MoNArch will use
the results of the project as an input to the Technology Plan, a document (internal
to Telecom Italia Group) that lists the available technologies and that is the basis
to develop the investment plan together with the strategy plan. Therefore, this will
have a direct impact on future investments and network operating costs, and could
be exploited to develop prototypes of new services. Moreover, the project will be
exploited by TIM by influencing the industry with requirements; the same applies
with the standardisation activity within the bodies participated by TIM. The over-
all activity within the project will be exploited as an opportunity to show to the
investor community the sensitivity of Telecom Italia towards innovation and fore-
runner research.
More specifically, TIM leads the demonstration activity in the Turin testbed
in Palazzo Madama, coordinating the partners’ involvement in the deployment of
the elasticity feature in network slicing. This activity has also an important aspect
related on the relationships established with the Municipality of Turin and the
“Fondazione Torino Musei” that were able, even if not partners in the consortium,
to exchange their requirements on the 5G deployment in the testbed.
METIS
METIS provided the basis for the all-communicating world envisioned beyond
2020 by proving a first investigation of technology concepts addressing the effi-
ciency, versatility and scalability of future communication systems. METIS was
able to determine which technology components in wireless communication sys-
tems may pursue an evolutionary path, and where complete paradigm changes
and revolutionary approaches were needed to meet the ambitious targets.
Through its strong consortium and the right timing, METIS was expected to
achieve a European-lead global consensus on fundamental questions connected
to the further development of wireless communication systems. In order to
achieve this, METIS pursued a two-fold research approach: Horizontal Top-
ics which enabled the future wireless communication system; and Technology
Components as being the foundations of the Horizontal Topics.
On one hand, METIS developed and enhanced the most promising technol-
ogy solutions/scenarios and use cases (called as Horizontal Topics) for be-
yond 2020 communications, such as: Direct Device-to-device Communication,
Massive Machine Communication, Ultra-dense Networks, Moving Networks,
and Ultra-reliable Communication. These technology scenarios and use cases
are illustrated in the figure and explained in the sequel.
Mauro Renato Boldi

TIM
Promising technology solutions and use cases for beyond 2020 communication
• Direct Device-to-device (D2D) Communication implies user-plane com-

munication directly between mobile devices, without going via any network
infra-structure. D2D communication offers the possibility for enhanced effi-
ciency and improved service quality in densely populated networks by re-
ducing the radio-link distances. The objective of METIS was to integrate
D2D as part of the overall communication network. Therefore, METIS stud-
ied adequate resource allocation and interference management solutions, al-
lowing the integration of D2D communications in the multi-dimensional
networks.
• Massive Machine Communication will be vital ingredients in future wire-
less communication networks. Communicating machines are predicted to be
an important source of traffic in the future all-communicating world. The
corresponding machine-related communication will be associated with a
wide range of characteristics and requirements that will in many cases devi-
ate substantially from the characteristics and requirements of the currently
dominating human-centric communication. The technologies researched and
METIS 327
developed within the METIS project addressed these characteristics and re-
quirements and provided enablers for efficient support of massive machine
communication within the future all-communicating world.
• The concept of Moving Networks enabled an efficient support of large pop-
ulations of jointly moving communication devices. The term “Moving Net-
works” refers to a number of novel concepts that focus on moving and/or
nomadic network nodes (e.g. vehicles, busses, trains, ferries, airplanes, etc.).
Note that a moving network node goes beyond the classical definition of a
moving relay as being discussed for LTE-A: a moving network node (e.g.
vehicles or busses with advanced communication and networking capabili-
ties) or a group of such nodes can form a “moving network” that communi-
cates with its environment i.e. other fixed or mobile nodes that are inside or
even outside the moving entity. The concept of moving and nomadic4 net-
work nodes relates to most of the technology components. The METIS ob-
jective for moving networks was to design wireless backhaul communica-
tion, mobility management such as interference and handover management,
as well as models for spectrum and network sharing.
• Ultra-dense Networks enables capacity, cost and energy efficiency: Infra-
structure densification is a path that has already been taken within e.g. exist-
ing cellular radio-access technologies, with inter-site densities as small as in
the order of 200 m in currently existing deployments. However, METIS
planned to take network densification even further, considering network-in-
fra-structure densities being order of magnitudes higher than today and
where at some point one may even have scenarios with, locally, a higher in-
fra-structure than device density. The aim was to enable extreme traffic ca-
pacity and service level. Ultra-dense deployments raise many new chal-
lenges e.g. related to mobility and backhauling. METIS provided/designed
the physical layer and network-layers functionally e.g. related to interference
management and mobility management for this enabler. Moreover, METIS
investigated the performance of ultra-dense networks in terms of cost, en-
ergy and spectral efficiency.
• Ultra-reliable Communication enabled availability and scalability: METIS
aimed at providing scalable and cost-efficient solutions for networks
4
The concept of nomadic network nodes defines a base station-like network node that
can be activated and deactivated and change its location in a flexible manner depend-
ing on traffic, service and coverage demands.
supporting services with extreme requirements on availability and reliabil-

ity. Herewith, METIS met the upcoming demand in domains such as Inter-
net of Things, telematics, automatization requiring wireless support. METIS
also aimed at providing concepts and solutions for the evolutions and migra-
tion of existing dependable and secure networks allowing the respected mar-
ket domains to benefit from the economy-of-scale of the public wireless
communication market.
On the other hand, METIS conducted research on the technology components
needed to support the technology solutions and use cases stated before. These
technology components were
• Air Interface Design, Waveforms and Multiple Access
• Multi-node coordination, multi-antenna, and multi-hop communications
• Heterogeneous multi-layer and multi radio access technology
• Frequency band analysis, scenarios, flexible spectrum access, coexistence
The two-fold research approach pursued in METIS led to a matrix structure of
the project. METIS integrated the horizontal topics as the project driving topics.
These enablers rely on the technology components (i.e. vertical dimension) e.g.
link, multiple access, multi-node, multiple-dimensional network and spectrum
aspects. The vertical dimension is crucial since this is how technology is dis-
sected and evolves, and how standardisations activities are classically organised.
While the vertical dimension allowed technology-driven innovation and more
efficient technical dissemination, the horizontal dimension ensured to capture
emerging market, societal and economical needs through the topic “Potential
Novel Scenarios”. In summary technology components were integrated as whole
and tightened together by emphasizing on the horizontal topics such as Direct
Device-2-Device Communication, Massive Machine Communication, Moving
Networks, Ultra-dense Network, Ultra-reliable Communication, and any new
promising enablers indentified during the course of the project time life. As such
the technology components pushed for achieving the needs of the HTs. In return
new HT will pull the technology toward new directions.
METIS 329
and
Frequency band analysis, flexible spectrum access, coexistence

Air Interface Design, Waveforms & Multiple
Frequency band analysis, flexible spectrum

multi--layer & multi radio access technology
Heterogeneous multi-layer & multi radio
multi--hop
multi--antenna,
Air Interface Design, Waveforms & Multiple Access
D2D Communications
and multi
communication
Massive Machine Communications
access, coexistence
access technology
multi--antenna,
multi
communication
Moving Networks
Access
Horizontal
coordination, multi
Topics
Ultra Dense Networks
-node coordination,
multi--hop
Heterogeneous multi
Ultra Reliable Communications
multi
-node
“Potential Novel Scenarios/Technologies”

Multi-
Multi
Multi-
Multi
Technology Components
The two-fold research approach pursued in METIS
Consequently, METIS addressed the 2020 information society needs and re-
quirements, and was able to attract new stakeholder segments and existing tele-
com industry. As stated, the technology components may provide the basis for
completely new wireless communication systems or may be integrated as part of
a long-term evolution of current wireless communication systems, with a target
market impact around year 2020 and beyond. The project outcome was to find
the modus operandi between potential new wireless communication solutions,
based on a preliminary assessment and a common understanding of the gain of
various potential concepts; and technology components for the evolution of
IMT-Advanced, which can be considered as retro-fit concepts to legacy stand-
ards.
TIM was the only Italian partner in METIS and this gave to the company a com-
petitive advantage in the race towards 5G, that was started practically from the
scratch by METIS project. The activity performed in the project was the basis
for all the future developments in 5G research in Europe and beyond. The
NGMN white paper on 5G, that many consider as the kick-off of the 5G era, was
largely inspired by METIS, and many partners were working in METIS and
NGMN. Also in the standardization process, the findings of METIS were of par-
amount importance; as an example, all the ITU categorization of use cases for
5G is based on METIS deliverables. The three well known families of use cases
(eMBB, mMTC, URLLC) are directly derived from the pioneering activity per-
formed in METIS.
METIS II
METIS-II is the project of 5G-PPP call for Phase 1 that is considered a direct
derivation of the project FP7 METIS (2012-2015) that laid the foundation of 5G
system in Europe.
METIS-II provided a 5G RAN design for an optimized support of wireless
communication services for industrial, public and private businesses. It is ex-
pected that the support of the variety of 5G use cases requires the existence of
multiple 5G air interfaces (or variants of them) that have different characteristics
in terms of coverage, latency, capacity, frequency bands, mobility, complexity,
deployments (indoor/outdoor) etc. In addition to this, it is also assumed that an
overall 5G system may benefit from these multiple air interfaces being more
closely integrated than in the case of 2G/3G/4G in order to fulfil novel require-
ments such as ultra-reliability, very high peak data rates, seamless mobility, etc.
Also, an important aspect will be the investigation on how other legacy air inter-
faces (e.g. LTE evolution and Wi-Fi) can be more efficiently integrated in the
5G domain. The most suitable tightness of integration among novel air interface
variants introduced in the 5G timeframe and legacy air interfaces will be deter-
mined during the project.
Mauro Renato Boldi

TIM
The METIS-II project built upon METIS and other projects related to 5G,
but went significantly beyond the achievements in these projects. In particular,
the METIS-II project pursued the following ambitious objectives:
1) Develop the overall 5G radio access network (RAN) design (where 5G
refers to the overall wireless communications system including evolved leg-
acy and novel radio access technologies), in a level of detail in between
“Technology Readiness Level 2” and “Technology Readiness Level 3”, and
focusing particularly on designing the technology for an efficient integra-
tion of legacy and novel radio access network concepts into one holistic
5G system.
2) Provide the 5G collaboration framework within 5G-PPP for a common
evaluation of 5G radio access network concepts from both a performance
and techno-economical perspective. More specifically, METIS-II further re-
fined 5G scenarios, requirements and KPIs, developed a performance and
techno-economical evaluation framework, and provided consolidation and
guidance to other 5G-PPP projects on spectrum and overall 5G radio ac-
cess network design aspects. Further, METIS-II developed an open-source
5G evaluation and visualisation tool for illustrating the key use cases of a
5G system as such, and the benefit of the key radio access network design
elements developed.
3) Prepare concerted action towards regulatory and standardisation bod-
ies for an efficient standardisation, development and economically attractive
roll-out of 5Gwith a strong European footprint and head start.
METIS-II had to overcome many unsolved technical problems by the following
key innovation pillars. These pillars are essential in the context of integrating
METIS II 333
legacy and novel air interface technologies into a holistic 5G radio access net-
work, and these inherit potential paradigm changes in cellular system design,
which are especially important to be addressed in wide industry consensus:
• Holistic spectrum management architecture. METIS-II defined, in col-
laboration with other projects and related forums, a framework and architec-
ture for flexible spectrum management and multi-operator collaboration in
5G. The key difference to METIS activities is significantly more focus on
the integration of spectrum beyond 6 GHz, consideration of “new 5G user
groups” and vertical industries as well as the increased flexibility needed to
make best use of different kinds of spectrum access mechanisms available.
In contrast to the past, not only the overall quantity of the available spectrum
but the different qualities of spectrum (bandwidth, propagation conditions
etc.) will define its potential usage and drive the work in METIS-II. The ini-
tial acceptance of spectrum sharing concepts like Licensed Shared Access
(LSA, such as in CEPT FM 52 and 53) opens the door for a broader adop-
tion of spectrum sharing concepts dynamic spectrum sharing, for other
bands as well. The outcome of WRC-15 and the preparation for WRC-18/19
opened new possibilities for spectrum usage to be exploited in 5G. Integrat-
ing different spectrum usage options (covering licensed and licence-ex-
empted bands) has been considered for network offloading but for vertical
industries using these bands.
• Holistic air interface harmonisation framework. METIS-II developed a
framework to harmonize similar protocol functionalities in the different pro-
tocol layers of air interface variants to be used in 5G, including both legacy
and new air interfaces, with the aim to keep device and infrastructure com-
plexity tractable and ensure a lean standard. This harmonisation can for in-
stance be performed by simply introducing common air interface character-
istics, such as frame structures, or functionalities, for instance on the PHY or
MAC layer, across multiple air interface variants. Alternatively, generic air
interface functionalities can be designed that can be parameterized to suit the
needs for different air interface variants. It is of course essential that harmo-
nisation is only applied to the extent that does not sacrifice the performance
of individual air interfaces too much. While there had already been previous
work on, e.g., developing novel air interfaces scalable to some extent of 5G
scenarios, there had yet been no work on a holistic air interface harmonisa-
tion framework considering all air interfaces, including novel and legacy
ones, needed to respond to the overall 5G requirement space.
• Agile Resource Management (RM) framework. The METIS-II project de-

fined a new complex multi-link, multi-layer network ecosystem. There, new
flavours of control and user plane integration and protocol adaptation were
present, which provided novel opportunities to re-design several functions
(e.g. interference management, power control, RAN moderation etc.) to
meet the 5G requirements. To achieve this, the project made additional ab-
stractions to the established concept of radio resource management (RRM)
and designed a holistic multi-air-interface framework for assigning services
to the most suitable resource not only in terms of time, frequency and space,
but also in terms of air interface variant and network processing resources
(i.e. real and virtual network elements). A key aspect of this framework is
that it enabled an easy integration of existing and novel air interface vari-
ants, and facilitated the introduction of potential novel air interface variants.
In this context, METIS-II investigated different design options, such as a
split of RRM functionality into an air-interface-dependent and air-interface-
agnostic part.
• Cross-layer and cross-air-interface system access and mobility frame-
work. The METIS-II project extended the cross-layer optimization concept
to another dimension, enabling a higher interaction of functionalities from
the multiple air interfaces in order to improve the overall resource usage.
New procedures were designed in such a way that functionalities of one air
interface could be used in a flexible way by another air interface, where this
coordination is dictated by a control plane framework which is common for
these air interfaces. To give an example, one of the concepts that were stud-
ied was a Unified System Access, where unified system information for all
the air interfaces is made available to the mobile terminals and possibly co-
ordinated random access actions may occur across the multiple air inter-
faces. Within this context, it was also investigated how the 5G system can
benefit from the existing infrastructure of legacy systems (e.g. LTE).
• Common control and user plane framework. METIS-II designed a com-
mon control/user plane framework for multiple 5G air interface variants, in-
cluding the evolution of legacy standards such as LTE-A. This will allow
the further development of multi-connectivity features such as fast link
switching, control/user plane diversity, throughput aggregation, etc. in order
to fulfil the different 5G requirements. To achieve this goal, integration al-
ternatives over the multiple protocol layers (RRC/PDCP/RLC/MAC/PHY)
METIS II 335
were investigated and a recommendation was issued in terms of which lay-

ers should be common to the different air interfaces based on different crite-
ria such as the required synchronicity between the different layers, end-to-
end delays, potential benefits from coordination, etc. Control/user plane
functions were evaluated as candidates for virtualization.
TIM was the only Italian partner in METIS-II, as a follow up of the participation
of TIM to the “parent” project METIS, that run from 2012 to 2015 in the context
of the call FP7 and laid the foundation to the research activity about 5G all over
Europe and beyond. METIS-II was based on a sub-set of partners that were also
partners of METIS and was intended as the “Target Action 1” in the 5G-PPP
pre-structuring model for Phase 1 projects. In this role, METIS-II concentrated
its activity on the overall end-to-end implementation of the 5G architecture and
most of the findings of the projects were the basis for the standardization activi-
ties leading to 3GPP Release 15 and partially also Release 16.
In the project, TIM, as well as contributing actively on many topics (spec-
trum, overall architecture definition, …), oversaw the standardization and dis-
semination Work Package. In this role, TIM promoted the activities of the pro-
ject, finalised in 2018 with the preparation of the book “5G System Design – Ar-
chitectural and Functional considerations and long term research” edited by
Wiley (Mauro Boldi main TIM author) and presenting a broad view on all the
most challenging topics in the implementation of 5G systems.
5G-CORAL
The 5G-PPP Phase 2 project termed “5G-CORAL: A 5G Convergent Virtualized

Radio Access Network Living at the Edge” aims at identifying and experimentally
proofing – through trials on large-scale testbeds – which are the key technology
innovations enabling the development of a virtualized Edge and Fog architecture
with convergence of multiple Radio Access Technologies (RATs) which is also
scalable, flexible and interoperable with other network domains including
transport, core network and distant Clouds. It is well known, in fact, that the 5G
system is expected to provide a wide range of services not limited to mobile broad-
band. These services, especially the ones which are bandwidth-intensive (e.g.
massive IoT) and/or delay-sensitive (e.g. V2X communications), require network-
ing, computing and storage capabilities to be provided closer to the end-users, i.e.
in the Edge of the mobile network. It is here where users’ data originates and
needs to be collected, processed and possibly sent to other nearby users, hence
leading to the concept of the Intelligent Edge. Techniques allowing for the above-
mentioned capabilities to be moved out of remote centralized Data Centres (i.e.
the Cloud) have been defined in the context of ETSI Multi-access Edge Compu-
ting (MEC) in order to meet 5G services’ stringent requirements. This approach,
however, cannot be implemented without considering the Network Functions Vir-
tualization (NFV) framework, also developed by ETSI, which allows both fixed
Mauro Renato Boldi

TIM
and mobile network functions to be software-implemented and then instantiated

by the operator on commodity off-the-shelf (COTS) servers by exploiting virtual-
ization technologies. The Edge computing approach is also needed for IoT appli-
cations due to the fact that there is the need to properly handle the significant
amount of data originated by “things” at the Edge, which negatively impacts mo-
bile network performance in terms of both required bandwidth and latency. This
issue can be solved by allowing data processing and applications to be concen-
trated locally in smart devices rather than being performed in the Cloud, leading
to the so-called Fog computing architectural paradigm. Actually Edge and Fog
approaches can allow not only (virtualized) RAN but also transport and core net-
work functions to be executed in the Edge and Fog tiers of the network so as to
save bandwidth in their respective domains and offer local breakout when needed.
Furthermore, in order to continuously offer wireless connectivity to various types
of devices and services in a given local area, a variety of RATs having different
characteristics in terms of capacity, spectral efficiency, mobility support, com-
plexity and costs needs to be integrated into a common framework, realizing the
so-called multi-RAT convergence. Integration of RAN functions belonging to
various RATs (e.g., LTE, WiFi, Narrow Band IoT, Bluetooth Low Energy,
ZigBee and so on) within an Edge and Fog virtualized architecture enables sim-
pler and cheaper networks to be realized via re-use of common functionalities and
infrastructure, also extending the service coverage of one RAT to areas where
other RATs are available without the need to deploy additional network equip-
ment. In addition, the possibility to dynamically select the best-suited RAT for
the service(s) the user is requesting helps in continuously ensuring the user’s
Quality-of-Experience (QoE). Another important aspect addressed in the Project
is the need to consider computing, storage and network resources being provided
by any Edge/Fog node or device on the move and regardless of the respective
owners. This goes beyond the ETSI MEC, where a fixed and centralized location
for the Edge platforms owned by only one stakeholder (i.e., the operator) is con-
sidered.
The 5G-CORAL concept is illustrated in Figure 1. It sets root in a hierarchical
multi-tier computing infrastructure, from Clouds and central Data Centres (DCs)
on top, down to Edge Data Centres (Edge DCs), and further distributed down into
Fog Computing Devices (Fog CDs) available locally in the access area. The focus
of the Project is on the Edge and Fog tiers of the distributed computing infrastruc-
ture, along with their interaction with the distant tiers. The concept comprises of
two major systems:
5G-CORAL 339
• the Edge and Fog computing System (EFS), as the host environment for the
virtualized functions, services, and applications
• the Orchestration and Control System (OCS) for managing and controlling
the EFS, including its interworking with other, external (i.e. non-EFS) do-
mains (e.g., transport network, core network, distant Clouds and DCs, etc.).
The 5G-CORAL solution will be validated in three real and large-scale testbeds,
identified and characterized according to the mobility requirement, i.e. low, me-
dium and high mobility. The testbeds, namely (i) Shopping Mall (low mobility
scenario), (ii) Connected Cars (medium mobility scenario) and (iii) High-Speed
Train (high mobility scenario) are located in Taiwan and, for the Connected Cars
only, also in Italy.
Figure 1 - 5G-CORAL concept.

The main objectives targeted by 5G-CORAL are the following:
Objective 1: Develop a system model that includes use cases, requirements, ar-
chitecture and business models to design and validate the 5G-CORAL solution.
This objective aims at identifying (and prioritizing) use cases, deployment sce-
narios and requirements for the design and demonstration of the 5G-CORAL so-
lution which leverages existing industrial frameworks for NFV, SDN, MEC and
Fog computing. The 5G-CORAL system framework also needs to support con-
vergence of multiple RATs. The business perspectives of the 5G-CORAL solution
will also be identified by considering all the stakeholders of the 5G-CORAL value
chain, e.g. operators, vendors, service/application/Cloud providers, facility own-
ers, end-users. Finally, step-based procedures and techniques for enabling incre-
mental deployment of the 5G-CORAL solution into existing networks will be de-
fined.
Objective 2: Design virtualised RAN functions, services, and applications for

hosting in the 5G-CORAL Edge and Fog computing System (EFS).
The virtualization of RAN functions on the EFS is explored for multiple RATs,
by developing their requirements and assessing their merits from an access con-
vergence viewpoint. In addition, EFS services will be specified for collection, ag-
gregation, publishing and use of radio and network context information by appli-
cations and possibly virtualised functions. EFS functions will be developed by
using EFS services from multiple RATs and the transport and core networks in
support of access convergence, e.g. multi-RAT convergence function to optimise
the traffic delivery across the multiple RATs. Finally, EFS applications will be
developed by using EFS services from multiple RATs and the transport and core
networks to improve network Key Performance Indicator (KPIs) and user QoE.
A proof of concept for the EFS will also be developed in order to demonstrate
EFS functions, services, and applications, with at least two coordinated RATs.
Selected EFS components will also be validated through a system verification in
the integration testbeds.
Objective 3: Design an Orchestration and Control system (OCS) for dynamic fed-
eration and optimised allocation of 5G-CORAL EFS resources.
Existing industrial frameworks for NFV, MEC, and Fog will be extended to best
suit dynamic environments where EFS resources are volatile. In addition, federa-
tion mechanisms for EFS resources belonging to multiple owners and subject to
different technical, business, and administrative requirements will be developed.
Also interfaces for automated deployment of EFS functions and applications will
be specified. Finally the integration of the EFS with central Clouds to enable in-
stantiation and migration of virtual functions and applications between the EFS
and central Clouds, together with the development of orchestration and control
algorithms for elastic placement and migration of EFS functions and optimised
allocation of EFS resources, are also other important parts of this objective.
5G-CORAL 341
A proof of concept for the OCS will be developed to prove and demonstrate:
(1) Federation mechanisms in static and dynamic scenarios; (2) Orchestration al-
gorithms aimed at reducing latency for EFS functions and applications; and (3)
Aggregation and offloading mechanisms for the best use of multiple RATs in the
EFS. Selected OCS components will also be validated through a system verifica-
tion in the integration testbeds.
Objective 4: Integrate and demonstrate 5G-CORAL technologies in large-scale

testbeds making use of facilities offered by Taiwan, and measure their Key Per-
formance Indicators (KPIs).
Existing testbeds in Taiwan will be customized in order to meet the needs of the
5G-CORAL proof-of-concept in large-scale deployments. This requires integra-
tion and validation of both EFS and OCS in large-scale testbeds (i.e. Shopping
Mall, High-Speed Train, and Connected Cars). Multi-RAT access convergence as
well as low latency applications (e.g. augmented reality and car safety) will be
demonstrated in real-world scenarios involving real users. The performance of the
5G-CORAL solution will be evaluated on-field via measurement of relevant KPIs
on data rates and latency in low and high mobility environments.
Objective 5: Disseminate and contribute 5G-CORAL results into international re-

search and innovation venues to pave the way for their successful exploitation.
This objective relates to the communication and dissemination of the 5G-CORAL
technical findings: the specification of a proactive standardisation plan (including
roadmaps, intellectual property creation, and contribution to relevant standards)
as well as an exploitation plan of the 5G-CORAL results into value creation for
all stakeholders during the project lifetime and beyond are parts of this objective.
Two Italian partners contributed to the Project: Telecom Italia (TI) and Azcom
Technology (Azcom).
TI had the role of Work Package 1 leader (and also task T1.1 leader) and was
focused on driving the use cases’ definition aligned with the envisioned network
evolution and business perspectives from a European operator’s point of view. TI
also collaborated with Azcom for the definition, setup and integration of the Con-
nected Cars testbed as well as to the experimentation and evaluation of the car
safety communication proof of concept. Initial experimental activities related to
the safety use case, in fact, have been conducted in TI premises in Torino (TILab).
Finally TI also contributed to disseminate the Project findings in public events
such as the IEEE 5G Summit held in Trento (Italy) in March 2018.
Azcom actively participated in WP1 for the definition of use cases, require-
ments, and business perspectives related to the 5G-CORAL car communications.
Furthermore, Azcom contributed to the specification of the EFS by focusing on
the Connected Cars’ functions and applications. In the context of the experimental
activities, Azcom lead task T4.1 and defined the Connected Cars testbed and
trialled the related use cases, such as the delivery of safety-related messages lev-
eraging the low-latency of 5G-CORAL solution. Finally, Azcom also contributed
to dissemination activities addressing the area of the vehicular communications.
5G-MiEdge
Current research efforts on 5G Radio Access Networks (RAN) strongly focus on

millimeter-wave (mmWave) access for addressing a critical weakness of deployed
cellular systems, i.e. the capacity to realize enhanced mobile broadband (eMBB)
services, as discussed at the World Radio-communication Conference 2015
(WRC-2015). Recently, mmWave technologies have reached a significant degree
of maturity and their state-of-the-art products, operated in the 60GHz unlicensed
band, are already in the market. However, there are many unsolved issues for an
effective deployment of mmWave 5G RAN. The most critical issue is the impos-
sibility of providing 10 Gigabit Ethernet backhaul everywhere. Meanwhile, Net-
work Function Virtualization (NFV) and Mobile Edge Computing (MEC) have
obtained much attention in 5G networks. Especially MEC is considered as a key
technology to enable mission-critical (low latency) applications by allocating stor-
age and computation resources at the edge of the network, so to circumvent the
backhaul networks’ limited capacity. However, in the case of mobile networks, it
is not easy to reallocate computational resources on demand, while meeting the
strict latency constraints foreseen in 5G networks. In this project, mmWave access
and MEC are firstly combined as a perfect couple of technologies compensating
each other’s weaknesses, and the resulting new system is called mmWave edge
cloud (MiEdge) henceforth. To facilitate MiEdge, the second main goal of this
project is to develop a new cellular network control plane in which context
Mauro Renato Boldi

TIM
information (location, traffic, classes of applications, etc.) of users are collected
and processed to forecast traffic and users’ requests, in order to enable a proactive
resource allocation. This novel control plane is called liquid RAN C-plane, since
the services and connections can follow and adapt to users’ need like a liquid to
the form of its container. Lastly, the users or application providers orchestrate
MiEdge to create a user/application centric 5G network that supports both eMBB
and mission-critical applications. In summary, the 5G MiEdge project develops a
feasible 5G ecosystem by combining MiEdge, liquid RAN C-plane, and user/ap-
plication centric orchestration. The project contributes to the standardization of
mmWave access and liquid RAN C-plane in both 3GPP and IEEE. At end, the
project demonstrates a joint 5G test-bed in the cities of Berlin and 2020 Tokyo
Olympic Stadium.
Figure 1-1: 5G MiEdge concept, use cases (left side) and key technologies (right side).
Objective 1: Research, develop, and prove the 5G based MiEdge concept

The project has the objective to research, develop and prove the 5G based
MiEdge concept whose viability is evaluated by detailed theoretical and numeri-
cal analysis and prototyped for proof-of-concept.
5G-MiEdge 345
Objective 2: Develop transmission schemes and protocols of mmWave ac-

cess/backhauling
The project develops transmission schemes and protocols of mmWave ac-
cess/backahauling aimed to assist the mobile edge cloud with caching/prefetching
so as to realize ultra-high speed and low latency service delivery, resilient to net-
work bottlenecks, such as e.g. backhaul congestion, users’ density, mission-criti-
cal service deployment, assuming three target scenarios: stadium, office, and
train/station.
Objective 3: Develop novel ultra-lean and inter-operable control signaling over

3GPP LTE
The project develops novel ulatra-lean and inter-operable control signaling over
3GPP LTE to provide liquid ubiquitous coverage in 5G networks based on acqui-
sition of context information and forecasting of traffic requirements, in order to
enable a proactive orchestration of communication/computation resources of the
mmWave edge cloud.
Objective 4: Develop user/application centric orchestration algorithms and pro-

tocols to adapt radio and computation resources of mmWave edge cloud in 5G
networks
The project develops user/application centric orchestration algorithms and proto-
cols to adapt radio and computation resources of mmWave edge cloud in 5G net-
works by utilizing traffic forecast provided by liquid RAN C-plane to enable self-
organized and proactive reservation of the resources and satisfy low-latency ser-
vice requirements.
Objective 5: Develop a joint 5G test-bed integrating mmWave edge cloud, liquid

RAN C-plane, and user/application centric orchestration
The project develops a joint 5G test-bed integrating mmWave edge cloud, liquid
RAN C-plane, and user/application centric orchestration to foster an effective im-
pact of 5G MiEdge in both Europe and Japan, particularly in preparation of 2020
Tokyo Olympic Games. The 5G MiEdge test-bed will liaise actively with the
other EU/JP consortium focusing on the network side as well as to leverage syn-
ergies between alternative 5G concepts.
Objective 6: Contribute to the definition of 5G mobile communications standards

in 3GPP and IEEE
The project contributes to the definition of 5G mobile communications standards
in 3GPP and IEEE as well as in open fora such as NGMN, Small Cell Forum, and
the International Telecommunication Union (ITU) Industry Specification Group
MEC, in terms of mmWave access, liquid RAN C-plane, and protocols for
user/application centric orchestration by coordination across European and Japa-
nese partners.
TIM, together with the University of Rome “La Sapienza” are the two Italian part-
ners in the 5G_MiEdge project consortium. Both TIM and University of Rome
“La Sapienza” are involved in the definition of use cases and reference scenarios
for the project activities. Concerning the research areas the University of Rome
“La Sapienza” is mainly involved in theoretical/simulative analysis concerning
dynamic resource allocation as well as more MEC focused investigations like pro-
cedures for joint optimization of radio computations resources under latency con-
straints. The acquired knowledge is disseminated to industry and society through
major international peer-reviewed conferences, workshops and journals and the
latest topics research results can be introduced in the education courses of students
at the university. At the same time, PhD and master students whose research topics
are related to 5G-MiEdge are involved in the project.
TIM, as industrial partner, has addressed its investigations toward more ex-
perimental activities in order to exploit the advanced 5G testing infrastructures
available in its Labs. TIM owns an Antenna lab designed for mmWave antenna
testing. The lab is based on a Compact Antenna Test Range (CATR) installed in
a shielded anechoic chamber designed to cover frequency range from 3 GHz to
110 GHz. Within 5G-MiEdge
TIM tests some of the mmWave antenna prototypes designed and developed
by the partners.
The project results on an extremely innovative subject like the synergic usage
of both mmWave access and MEC are very important to better understand how to
design and operate a network in the mmWave band able to provide eMBB services
with a very low latency constraint like, e.g., immersive reality applications.
5G Picture
5G-PICTURE proposes a paradigm shift, from the traditional RAN and recent C-
RAN to the “Dis-Aggregated RAN” (DA-RAN) approach. DA-RAN is a novel
concept adopting the notion of “disaggregation” of HW and SW components
across the wireless, optical and compute/storage domains. “Resource disaggrega-
tion” allows decoupling of HW and SW components creating a common “pool of
resources” that can be independently selected and allocated on demand. These
HW and SW components form the basic set of building blocks that, in principle,
can be independently combined to compose any infrastructure service. Apart from
increased flexibility, disaggregation, due to its modular approach, offers enhanced
scalability, upgradability and sustainability potential that are particularly relevant
to 5G environments supporting enormous and continuously growing number of
end-devices and services. To exploit the concept of disaggregation in RAN envi-
ronments, novel 5G technology solutions are needed to increase the density and
power efficiency of the “pool of resources”, supporting at the same time high
bandwidth connectivity between them. These will rely on
• hardware programmability: allowing HW repurposing to enable dynamic on
demand sharing of resources and
Mauro Renato Boldi

TIM
• network softwarisation: enabling migration from the traditional closed net-

working model, focusing on network entities, to an open reference platform
instantiating a variety of network functions.
Such novel networking approaches can facilitate increased functionality and flex-
ibility infrastructures, offering simplified management and advanced capabilities
including slicing and virtualisation that allow the
disaggregated resource pool to be shared and accessed remotely. On-demand
selection and allocation of these resources (flexible mix-and-match) will enable
provisioning of any service without the prerequisite of owning and installing any
specific HW or SW, adopting novel approaches such as the notion of service
chaining (SC).
Network slicing and service chaining will be facilitated adopting architectural
models such as the SDN reference architecture and the ETSI NFV standard.
Objective 1: Design and prototype an open 5G infrastructure capable of instanti-

ating and co-hosting different vertical sectors. This will be based on leading in-
dustry and open source technologies supporting very diverse service require-
ments.
The 5G-PICTURE SW/HW disaggregated network approach is key for the crea-
tion of a 5G infrastructure able to support a large variety of 5G ICT and "vertical"
services. According to the proposed solution, vertical service providers, currently
relying on closed and proprietary infrastructures, is able to deploy any service
without having to own and install any HW or SW component. The 5GPICTURE
solution allows end-users and third parties to access real or virtual equipment,
services, systems and tools on demand regardless of their geographical location.
This solution is expected to support any type of service ranging from delay sensi-
tive video, to infotainment services, and from best effort applications to ultrareli-
able ones such as M2M communications. This will enable transformation of ver-
tical sectors from closed inflexible environments into a pool of modular HW and
SW components that can be combined on demand.
Objective 2: enable a common, flexible and open underlying network infrastruc-
ture that can unleash market innovation and support wide adoption and new busi-
ness models and opportunities to a wider ecosystem (ICT and verticals) 5G-PIC-
TURE will adopt and develop HW technology innovations.
5G Picture 349
5G-PICTURE designs, develops and prototypes advanced and highly configura-

ble optical/radio infrastructure technologies and interfaces at the data plane, inte-
grated in a single transport solution supporting both 5G ICT and vertical opera-
tional and end-user services. In addition, 5G-PICTURE will develop advanced
solutions for programmability of data plane functionalities offering HW acceler-
ation for high performance and low latency processing, HW programmable net-
work interfaces, and platform independent programming abstractions. FH and BH
processing, together with the relevant exposed interfaces and programming mod-
els, including abstractions of existing HW, will be also supported. The upper lay-
ers will be provided with a stack of open SW and Application Programming In-
terfaces (APIs) coupled with the necessary metadata definition. These will enable
the abstraction and exploitation of the underlying HW platform taking into ac-
count specific requirements for closed platforms.
Objective 3: the integration of the SDN and NFV approaches and adopting the
concept of network functions that are functional building blocks with well-defined
external interfaces and behaviour.
The implementation of 5G RAN functional splits is determined by the underlying
platforms.
Given the heterogeneous compute and network infrastructure envisioned in
future 5G networks, 5GPICTURE investigates and experimentally demonstrate
optimal 5G functional splits for different types of infrastructure (compute and net-
work). It will also demonstrate the concept of flexible RAN functional splits that
are “adaptive” RAN implementations requiring platforms able to instantiate more
than one functional split. Flexible RAN functional splits are an enabler for slicing.
Furthermore, realizing that nowadays synchronisation protocols are based on
packet round trip time measurements that are subject to congestion, especially in
the wireless BH, 5G-PICTURE investigates novel approaches, including the PHY
layer, to deliver high accuracy synchronisation to heterogeneous wireless/opti-
cal/packet domains.
Objective 4: provide an operating system able to simplify the infrastructure man-

agement and the provisioning of a large variety of services that are complex con-
structs of a diverse set of functions, supported by the heterogeneous physical in-
frastructure and technologies.
The core value proposition of 5G-PICTURE is “HW programmability”: HW can
be configured to execute given functions exceeding the speed of SW running on
commodity hardware blurring the boundary between SDN (configurable packet

handling) and NFV (deploy functions into the network). 5GPICTURE extends the
concepts of SDN control and NFV orchestration and merge or, at least, tightly
integrate them. In addition, the role of a “slicing system” and its integration with
SDN/NFV is not obvious and has to be solved. A core contribution of this first
objective will be a set of interfaces between these system components as well as
advanced interfaces (e.g., one to feed predictions into the system).
Objective 5: Demonstrate converged FH and BH services in: a smart city envi-
ronment, the first 5G railway experimental testbed showcasing seamless service
provisioning and high speed mobility management, and a stadium supporting me-
dia services for ultra-high density users
The 5G-PICTURE solution enables the overall 5G vision, supporting any service,
including operational and end-user services for both ICT and “vertical" industries.
Proof of concept demonstrators will be showcased in realistic environments in-
cluding:
(a) 5G-railway testbed located in Barcelona, Spain comprising three tracks cov-
ering scenarios with the rolling stock. This will be the first 5G railway exper-
imental testbed to showcase support of seamless service provisioning and mo-
bility management in high speed moving environments,
(b) 5G-stadium testbed located in Bristol, UK to address scenarios with in-
creased density and static-to-low mobility. In this environment media services
associated with large venues will be demonstrated
(c) 5G-smart city testbed to experimentally validate the DA-RAN concept
through the support of joint BH and FH services. This test-bed will be sup-
ported and hosted by the state-of-the-art 5G “City of Bristol” network infra-
structure (BIO).
TIM and CNIT are the only Italian partner in 5G-PICTURE.

In the project, CNIT, involved with the University of Rome Tor Vergata re-
search unit, leverages, and significantly extends, data plane programmability
know how acquired (and technologies developed) in two previous projects: the
5G Picture 351
5G H2020 project SUPERFLUIDITY on network programmability, and the ICT-

05 BEBA on stateful network processing. CNIT will bring from SUPERFLUID-
ITY the work done in terms of functional decomposition of flow/packet-pro-
cessing tasks, including fine-grained decomposition of selected functions into
very elementary events/actions. From this decomposition a preliminary processor
architecture for packet manipulation has been identified. From the BEBA project
CNIT will leverage the stateful forwarding switch architectures including a first
small-scale OpenState-FPGA prototype (Mealy Machine execution on FPGA), its
corresponding software open-source implementation in OFsoftswitch13 with the
relevant Ryu controller extensions.
Within the project, TIM works on the definition of vertical systems and con-
sequent requirements and the architecture of integrated fronthauling and back-
hauling network . Also some details within the control umbrella will be touched.
Due to experience in managing EU projects and technical skills, Telecom Italia
has been appointed for the management and coordinate n of a workpackage de-
voted to the definition of the global architecture.
Fantastic-5G
FANTASTIC-5G (Flexible Air iNTerfAce for Scalable service delivery wiThin

wIreless Communication networks of the 5th Generation) targets to design a sin-
gle modular Air Interface (AI) allowing the concurrent support of this wide range
of use cases. This is deemed as essential because future communication systems
are not only to deliver Mobile BroadBand (MBB) services but also to support
Massive Machine Communications (MMC) and Mission Critical Communica-
tions (MCC) efficiently, therefore it is expected operators to be able to capitalize
on new revenue streams. Similarly, supporting Broad- and Multicasting Services
(BMS) in 5G will help widen the market and potentially free parts of the spectrum.
Finally, Vehicle-to-Vehicle and Vehicle-to-Infrastructure communications (V2X)
will also play a crucial role in future mobile networks, to more efficiently support
mobile services while on the move and to enable new use cases such as traffic
safety.
These above-mentioned five services, i.e. MBB, MMC, MCC, BMS and V2X
– also labelled by FANTASTIC-5G as “core services” in the 5G ecosystem – are
each linked to different sets of Key Performance Indicators (KPIs) to be opti-
mized, in terms of which the service requirements are defined. A simple and
straightforward solution to provide each service type is the design of a dedicated,
service-specific AI, which however represents an uncompetitive and unprofitable
Mauro Renato Boldi

TIM
approach when compared to the design of a single AI being able to be parameter-

ized in different ways to concurrently accommodate all use cases on a per-need
basis. According to this FANTASTIC-5G targets a single modular AI concur-
rently supporting a wide range of use cases: this cannot be achieved by following
a “one-fits-all” approach but, instead, by designing the AI in such a way that dif-
ferent parts of the AI and related functionalities can be properly parameterized in
order to satisfy the needs of the respective service being targeted.
FANTASTIC-5G is set out to design the 5G AI and proposes inputs for standard-
ization, whose timeframe is expected to start during the project lifetime. Five
trends which are pushing the limits of 4G have been identified, pointing towards
a need for the 5th generation cellular technology:
• Trend 1 – Increase in capacity: the demand for wireless data is predicted to
increase significantly, resulting in 1000x higher mobile data volumes and 10-
100x higher end-user data rates
• Trend 2 – Increase in the number of connected devices: the number of con-
nected devices is predicted to increase by a factor of 10-100, which means
that up to 300.000 devices need to be served per access point
• Trend 3 – Increase in reliability: wireless connectivity will be applied to new
use cases that require extremely reliable connections (typically 99.999%
availability) and mission-critical communications, such as Vehicle-to-Vehi-
cle coordination, critical control of the power grid, etc
• Trend 4 – Decrease in latency: remote presence and tactile Internet that im-
pose stringent latency constraints on the end-to-end connection, including the
wireless part. Forecasts indicate that the latency should be decreased by a fac-
tor of 5 in order to enable such services
• Trend 5 – Increase in efficiency: efficiency in terms of resource utilization
(e.g. energy, spectrum) is becoming more and more pronounced. It is seen as
an indispensable ingredient for a healthy/sustainable ICT market/business and
environment.
FANTASTIC-5G sets out a vision on a new 5G AI below 6GHz with the follow-
ing Key Characteristics:
• flexibility to support a broad class of services with their associated (broad
class) of KPIs
• scalability to support a high number of devices being connected
• versatility to support the diverse device types and traffic/transmission charac-
teristics
Fantastic-5G 355
• efficiency to support the requirements on energy consumption and resource

utilisation
• future-proofness to support easy integration of new features.
In order to make this vision a reality, FANTASTIC-5G is committed to conduct
work towards the achievement of the following high-level measurable objectives:
• Objective 1: to develop a highly flexible, versatile and scalable AI to enable
the in-band coexistence of highly differing services, device types and traf-
fic/transmission characteristics
• Objective 2: to design an AI enabling ubiquitous coverage and high capacity
where and when required
• Objective 3: to develop an AI being highly efficient in terms of energy and
resource consumption
• Objective 4: to render 5G more future-proof than former generations through
easier introduction of new features
• Objective 5: to evaluate and validate the developed concepts by means of sys-
tem level simulations and hardware proof of concepts for selected compo-
nents
• Objective 6: to build up consensus on reasonable options for 5G standardiza-
tion among the major industrial partners of the project that are also voting
members in 3GPP (Alcatel Lucent, Orange, Huawei, Nokia, Intel, Samsung
Electronics, Telecom Italia) and to push the innovations of the project for
standardization (through study items).
Figure 1 depicts how the first four (technical) objectives of FANTASTIC-5G are
linked to its vision and to the trends presented above. Note that in this figure there
is not a one-to-one correspondence between the trends, the key characteristics and
the objectives, e.g. Objective 1 is not an outcome of only the Key Characteristic
1 (Flexibility), which itself is not an outcome of only Trend 1. All objectives are
derived from the combinations of the Key Characteristics, which are the outcomes
of the ensemble of the Trends.
Figure 1: From trends to the FANTASTIC-5G objectives.
Essentially the FANTASTIC-5G research project is targeting at pre-standard-

ization development of a new 5G AI to influence future 3GPP 5G standardization.
By fulfilling the project objectives represents an interesting set of research chal-
lenges on how to most efficiently adapt the design of AI components (e.g. wave-
form, frame structure, channel coding, modulation, etc.) and related physical layer
procedures (e.g. support of MIMO, HARQ retransmissions, channel estimation,
synchronization, etc.) to the requirements of the multiple sources of heterogeneity.
The FANTASTIC-5G hypothesis is that this is possible through re-configuration
of flexible solutions for the AI components and related procedures.
Two important Italian partners contributed to the project: Telecom Italia (TI) and
the research group belonging to the Telematics Lab - Politecnico di Bari
(POLIBA).
Fantastic-5G 357
TI, being a key player in the mobile industry, was essential for the objective
of building up consensus on reasonable options for 5G standardization. As an op-
erator, TI was crucial for the selection of relevant scenarios, the identification of
requirements to be taken into account by the AI design and for the overall AI
design itself. Telecom Italia was involved in Work Package 2 (WP2), where it
was leading the task responsible for requirements definition in the Project (T2.1–
KPIs, requirements, and air interface framework). In addition, within WP2, TI
also contributed to Task 2.2 (System Design) and Task 2.3 (System Evaluation,
in conjunction with POLIBA). TI also had an effort in both WP6 tasks for Pro-
ject’s dissemination and exploitation.
TI used the results from FANTASTIC-5G as an input to the Group Technol-
ogy Plan, an internal document that lists the available technologies and that is the
basis to develop the Investment Plan together with the Strategy Plan. Moreover,
the project has been exploited by TI for influencing the industry with require-
ments, mainly focusing on:
• enhancing service quality for mobile broadband (in particular by increasing
UE throughput in hotspots, such as stadiums, where large concentrations of
people for short periods of time are expected), directly impacting customer
satisfaction, revenues and customer retention
• improving energy efficiency and general sustainability indicators
• exploiting Internet-of-Things (IoT) and, more in general, Machine-to-Ma-
chine communications as well.
All the above-mentioned areas of activity have a direct impact on future invest-
ments and network operating costs. The standardization activity within the project
has also been exploited by TI by ensuring that standards do take into account the
requirements of TI itself, in particular related to user throughput and to ensure a
proper Quality of Experience to the customer. The overall activity within the Pro-
ject also represented an opportunity to show to the investor community the sensi-
tivity of TI towards innovation and forerunner research. TI disseminated the Pro-
ject’s results via publication of articles in the TI magazines distributed within the
Group and available for experts in Telecommunication field but also by organiz-
ing workshops with national and international scope (including South America
subsidiaries).
The research group belonging to the Telematics Lab - POLIBA has a long
tradition of research on Quality-of-Service/Quality-of-Experience (QoS/QoE)
management in wireless systems, congestion control, wireless sensor network,
IoT, industrial protocol stacks, information centric networking, smart city and
smart grid, eco-sustainable heterogeneous networks, and system-level simulators
in wireless and cellular networks. In addition, it founded (and it is maintaining)
the open source project LTE-Sim, which is an open source network simulator tool
able to model a very large number of aspects belonging to both LTE and LTE-A
specifications. The network simulator tool was of high relevance for the Project:
promising technical component conceived in the context of the FANTASTIC-5G
project, in fact, have been implemented into the simulator for assessing system-
level gains.
Specifically, POLIBA was involved in WP2, contributing to both tasks T2.2
(System Design) and T2.3 (System Evaluation). In this context, POLIBA devel-
oped a flexible simulation tool being able to support the system-level evaluation
of WP2 technical findings. The LTE-Sim simulator was properly extended to in-
tegrate 5G environmental models (scenarios and services), abstraction of PHY
and MAC layers, and tools for the evaluation of KPIs. Moreover, to ensure right
and repeatable results, an accurate calibration of both channel and network mod-
els, that were available within the simulator, was carefully executed.
Moreover, POLIBA was involved in dissemination and exploitation activities
envisaged for the WP6. Besides disseminating research findings in conferences
with recognized international level and journals with high impact factor in the
field of communication networks, POLIBA included activities related to the FAN-
TASTIC-5G project within Masters and Doctoral courses, also promoting the dif-
fusion of the simulation platform in both academia and industry environments
being interested in 5G technologies (distributing it as an open source tool). In this
manner, several aspects of the project have been disseminated in a very broad
fashion.
METRO HAUL
METRO-HAUL involves the design and development of a novel, spectrally effi-

cient and adaptive network solution, using dynamic elastic optical networking,
including both transparent and flexible optical switching and adaptive transmis-
sion. METRO-HAUL will address the granularity mismatch between the wireless
access and the optical metro domain via a new edge node design (Figure 1), and
achieved through dynamic optical bandwidth allocation. This will provide metro
support for increased volume of services with reduced cost and energy consump-
tion.
To support the required dynamicity and flexibility, the METRO-HAUL archi-
tecture will be developed to integrate a wide range of optical technologies. These
will be controlled using automation schemes and programmability features that
will enable concepts such as HW disaggregation and virtualization, the coordina-
tion of which will be supported by a purposely designed control plane. Optical
nodes will be dynamically adapted to the needs of specific services, optimally
exploiting the data plane through use of relevant data monitoring and analysis
schemes.
The control plane will be also responsible for the provisioning of 5G and ver-
tical industry services that require the allocation of heterogeneous compute, stor-
age and networking resources and ensure the required end-to-end QoS and QoE
Mauro Renato Boldi

TIM
levels for each application. The METRO-HAUL control plane will leverage on
the well-established SDN and NFV paradigms and exploit the benefits of a unified
system, coordinating networking, computing, storage, transmission, and switch-
ing aspects, enabling abstraction at different levels.
Through the combination of improved, elastic-based optical techniques and
intelligent, dynamic management of 5G applications, METRO-HAUL will sup-
port more than100 times the capacity of current metro networks.
METRO-HAUL will perform PoC demonstrations to illustrate its network so-
lution, metro node prototypes, novel optical transmission technologies and asso-
ciated control plane / orchestration software. Detailed performance evaluation of
both data and control planes will be carried out. The final project demonstrations
will also involve the demonstration of real 5G and vertical services across the
METRO-HAUL test-bed. Finally, the project will actively participate in the rele-
vant standardization bodies to promote METRO-HAUL solutions to the wider
community.
Figure 1: Metro Network Supporting RANs, with distributed DC and virtualized / pooled
BBUs and EPC
Objective 1: To architect and design cost-effective, energy-efficient, agile and

programmable metro networks that are scalable for 5G access and future require-
ments, encompassing the design of all-optical metro nodes (including full compute
and storage capabilities), which interface effectively with both 5G access and
multi-Tbit/s elastic core networks.
The main objective is to improve metropolitan networks by defining an architec-
ture that supports 5G access requirements, notably in terms of low latency, mo-
bility, increased bandwidth and also dynamic storage and compute functions,
METRO HAUL 361
implying highly energy efficient operation. Given the scale of targeted deploy-
ments, metro networks must be cost-effective, including both CAPEX and OPEX.
This involves new node and transmission technologies, but also introduces com-
moditization of multi-vendor hardware leading to new business models
Objective 2: Establish the most cost and energy-efficient, high bandwidth, flexible
optical metro transport technologies to support continued exponential consumer
and business data demands together with the backhaul (or ‘METRO-HAUL’) of
5G traffic.
METRO-HAUL will explore a wide range of transport options that provide flex-
ible and cost effective high bandwidth, ranging from fully filterless to optical filter
based solutions
Objective 3: To explore and identify pros/cons of different multi-technology,

multi-vendor or disaggregated “whiteboxes” ecosystems for metro nodes and op-
tical metro-wide networks, and the impact on Operators’ O&M processes and
System Vendors’ business models.
The metro network’s reduced geographical range relaxes the optical performance
parameters, making it easier to integrate components from multiple vendors. O3
will find the aspects of METRO-HAUL that can be disaggregated, utilizing
whitebox components without compromising overall network performance, but
radically reducing cost. Several possible scenarios will be examined from both the
Operators’ and system Vendors’ points of view.
Objective 4: To design, implement and evaluate prototype Optical Infrastructure

Elements (OIE) to enable METRO-HAUL concept demonstrations..
The project will explore a range of optical components, including ROADMs and
other low-cost optical switching technologies, flexible transceivers and digital
switches. O4 will explore how to obtain high-capacity and flexibility in the eco-
nomically challenged metro environment.
Objective 5: To design and implement an SDN/NFV control framework and ar-

chitecture for heterogeneous metro-haul networks supporting 5G operational and
both end-user and vertical oriented services, including slicing
The objective targets the design and implementation of an augmented SDN/NFV
control framework and architecture specially adapted for METRO-HAUL and en-
abling the dynamic deployment of 5G services (i.e. the operation of both end-user
and corporate-oriented services). These services have requirements for resources

(processing, storage and networking) that change over time considering aspects
of the service life-cycle and variations in usage; the architecture will inherently
support multi-tenancy, network virtualization, and slicing
Objective 6: To build integrated prototype metro network demonstrators to show

the improvements to the performance of vertical 5G applications
The project intends to bring together the multiple innovations in the form of de-
monstrators – integrated Proof of Concepts designed to show key data plane and
control plane features and also designed to show how METRO-HAUL supports
5G Vertical applications
Objective 7: To design algorithms to improve efficient resource usage for the

overall METRO-HAUL solution
This objective targets the design and validation of algorithms to analyse and con-
trol different network aspects, focusing on optimization of the wide range of het-
erogeneous (network and IT) resources. Algorithms will cover computation
across the infrastructure, for the supported services, including dynamic metro re-
source activation, spectrum management, data analytics for Knowledge Discov-
ery from Data (KDD), network re-optimization and function placement.
Several partners are involved in Metro-Haul: TIM, CNIT, Coritel, Ericsson and
Nokia (former Alcatel Lucent) .
As an incumbent carrier TIM contributes with the definition of service re-
quirements including KPI and SLA, techno-economic feasibility studies and po-
tential migration roadmap identification. Furthermore, it will contribute to the def-
inition of Metro Node architecture to virtualize at once services and network func-
tionalities; to the separation of control and management planes from data planes;
and to the disaggregation of complex optical network elements in elementary
bricks to arrange as needed in a multivendor environment; to the development of
agents for optical white box control and management; and to the definition of final
demonstrators of project proposed solutions.
METRO HAUL 363
TIM feeds the Standardization bodies with Metro Haul outcomes.

CNIT contributes to METRO-HAUL through two distinct aspects. The first
aspect concerns the data plane, with particular focuses on: i) investigating solu-
tions for disaggregation of optical functions in a multi-vendor environment (opti-
cal white boxes), including both filtered and filterless scenarios, ii) the use of pho-
tonic integrated components (e.g. mini-ROADMs) as switching elements in opti-
cal white boxes, iii) flexible and programmable transceivers for high capacity
metro segment, providing advanced software-defined features and addressing
both coherent and direct detection.
CNIT also intensely works on the control plane for the METRO-HAUL ar-
chitecture. In particular, it will contribute to: i) the study, design and implemen-
tation of a Common Optical Node Control (CONC) SW for optical white boxes
as well as interfaces towards monitoring systems for optical resources and trans-
mission QoT, ii) the definition of functional requirements for control and pro-
grammability of the underlying metro-haul infrastructure, also focusing on moni-
toring and recovery aspects, iii) the definition of data models of the elements and
devices and the control of photonic-integrated components and node solutions for
enhanced filterless/semi-filterless scenarios, iv) to Service Function Chaining
(SFC), including algorithmic aspects for function placement and network virtual-
ization.
Ericsson Telecomunicazioni contributes to the project analysing the Metro
Access network segment and its evolution influenced by the evolving Fronthaul
and Backhaul networks for 5G Wirless Radio Mobile and Wireline access, by
defining the flexible deployment and the reconfiguration of network elements and
networking functions. The metro network segment may evolve to operate as the
functional backhaul for 5G network: Metrohaul. The network connectivity will be
designed with a unified network model, supporting different underlying data
planes and protocol schemes in order to better serve/recognize the end user appli-
cations/demands. Analysis is performed on the sustainability of a possible data-
com-telecom convergence solution in metro-access network enabling 5G: Work
is done on the design of an updated metro-access network with the lowest possible
impact in terms of cost, power dissipation and system complexity, enabling soft-
ware driven dataplane, network function virtualization, network automation and
improving technical performance. Interoperability issue based on white box con-
text will be addressed for instance by ensuring high quality of service and reliable
fault management. Contribution is given in the definition and demonstration of
relevant use cases supported by datacom-telecom converged network in 5G sce-

nario.
CoRiTel is involved in the design and evaluation of the proposed architecture
and it works on the control plane requirements, in the identification of data models
and the implementation of algorithms for traffic-based resource activation, and on
VNF placement algorithms. CoRiTel is also involved in WP1 in project organi-
zation and management, and in WP6 will contribute in dissemination aspects,
mostly related to scientific publication.
Finally, Nokia Italy contributes to METRO-HAUL with its expertise on soft-
ware architecture to design and provide a new SDN app for the management of
optical defragmentation in a multilayer network.
5G XCAST
5G-Xcast devises, assesses and demonstrates large scale immersive media deliv-
ery by means of conceptually novel wireless technologies, contributing to the fur-
ther definition of 5G and its standardisation. 5G-Xcast is focussing on large scale
media distribution, as this use case is one of the most demanding requirements in
terms of data rate (capacity), scalability (cost-effectiveness) and ubiquity (cover-
age). The 5G-Xcast media delivery solution will have built-in unicast/mul-
ticast/broadcast modes and caching capabilities, holistically approaching media
delivery harmonisation across the three communication modes with seamless user
experience, as well as common Application Programming Interfaces (APIs) to
content service providers. The project covers multiple disciplines from the radio
interface to the transport and application layers, including protocols and APIs, as
well as network and system architecture aspects. The development of the 5G-
Xcast media delivery solution is focused on the Media and Entertainment (M&E)
vertical. The automotive, Internet of Things and public safety verticals will be
considered to ensure that their respective technical requirements are also identi-
fied and suitably addressed. Proof-of-concept prototypes and technology demon-
strators are pivotal tasks of the project. Special emphasis is being given to emerg-
ing new immersive media services that cannot be efficiently delivered by existing
technologies and networks. 5G-Xcast will enable progressive convergence in the
media delivery sector while optimising the resources of the available infrastruc-
tures, enabling new business models over fixed, mobile and broadcast networks
that will be also investigated in the project. The consortium is a balanced combi-
nation of telecom and media entities, covering the complete M&E value chain,
Mauro Renato Boldi

TIM
with a very strong presence in 5G-PPP phase-1 and 3GPP standardization of 4G

Broadcast.
Figure 1-1: An example of one user moving between three different environments (identified by
the labels 1,2 and 3) each of which has different combinations of networks available. The converged
media delivery architecture of 5G-Xcast over fixed broadband, mobile broadband and terrestrial
broadcast networks allows a seamless, uninterrupted service to be offered to the users as they move.
Objective 1: To develop broadcast and multicast point to multipoint (PTM) capa-

bilities for 5G and evaluate 5G spectrum allocation options for PTM network de-
ployment
Broadcast is considered as an integral part of the 5G system from the beginning.
Multicast and broadcast appear as part of the 5G vision of many stakeholders,
including the 5G-PPP (see e.g. its vision, M&E whitepaper, and 5G vertical sec-
tors whitepapers). 3GPP has also identified broadcast transmissions as a critical
technical requirement for eight 5G use cases including, connected vehicles, in-
network and device caching, and broadcast/multicast services using a dedicated
radio carrier. The broadcast/multicast use case was also identified by the Next-
Generation Mobile Network (NGMN) alliance in, where the possibility that 5G
5G XCAST 367
could complement or even act as a substitute for the legacy radio/TV broadcast
networks was considered. However, this topic has received very little attention by
the 5G research community during phase-1 of 5G-PPP. Furthermore, 3GPP de-
cided to prioritize the work in other areas for Release (Rel)’15 and hence 5G
broadcast would only be addressed starting from Rel’16. This creates an oppor-
tunity for 5G-Xcast to develop novel solutions and contribute to the standardisa-
tion work in time for Rel’16 whose stage III will be frozen in Dec. 2019 allowing
for a maximum impact by the project.
5G-Xcast will provide ‘5G Broadcast’, an innovative evolution of MBMS
(Multimedia Broadcast Multicast Services). Both 3GPP RAN (Radio Access Net-
work) and SA (Service and System Aspects) aspects will be covered. 5G-Xcast
will operate a paradigm shift from 3G and 4G LTE (Long Term Evolution) net-
works, where the broadcast transmissions have been built as an add-on service.
The 5G-Xcast design will integrate 5G Broadcast as a built-in feature for delivery
optimisation. The proposed 5G broadcast air interface will outperform existing
wireless multicast/broadcast technologies in terms of spectral efficiency, through-
put, power consumption, latency and interactivity, enhancing the 5G Key Perfor-
mance Indicators (KPIs) relevant and applicable to the specific 5G-Xcast use
cases and scenarios, as described, for example, in Rec. ITU-R M.2083. 5G-Xcast
will also evaluate conceivable spectrum options with regards to their suitability
for 5G Broadcast2.
The project aims at making results available to e.g., CEPT (European Confer-
ence of Postal and Telecommunications Administrations), RSPG (Radio Spec-
trum Policy Group), and ITU (International Telecommunication Union) either di-
rectly or through 5G-PPP and initiatives contributing to international technical
work also contributing to the ITU-R IMT-2020 evaluation process.
Objective 2: To design a dynamically adaptable 5G network architecture with

layer-independent network interfaces capable of dynamically and seamlessly
switching between unicast, multicast and broadcast modes or use them in parallel
and exploiting built-in caching capabilities
Future 5G networks must support many very different use cases within one com-
mon technical framework. Clearly structured and transparent interfaces between
content service providers and network operators, as well as between different net-
work layers (radio, transport and application layers) are crucial to fulfil the
required tasks. The 5G-Xcast project will introduce and develop the concepts of
radio network-related elements (Radio Access Network, RAN), network-
aware elements (core network) and network-agnostic elements (content dis-

tribution framework). This will enable progressive convergence of media deliv-
ery system allowing soft switchover between technologies and infrastructure
whilst also enabling optimised resource utilisation of all available infrastructures.
5G-Xcast will enable the full integration of unicast, multicast and broadcast
modes under one common framework. Dynamical seamless switching between
different modes of operations, as well as their parallel usage, are essential to ex-
ploit network and spectrum resources in an efficient and economic manner. In
combination with advanced caching technologies, this approach will open a door
to completely new levels of network management and delivery cost-efficiency.
5G-Xcast will provide the necessary mechanisms.
Objective 3: To experimentally demonstrate the 5G key innovations developed in

the project for the M&E and PWS verticals
The 5G-Xcast consortium will experimentally test and validate the technical so-
lutions developed in the project for the M&E vertical. For that purpose, the pro-
ject will utilize three test-beds, including two urban city test-beds in Munich
(Germany) and Turku (Finland) from the two most relevant national research pro-
jects in Europe about 4G Broadcast, and a university campus test-bed of the 5G
Innovation Centre (5GIC) in Surrey (UK). The project aims to demonstrate the
key innovations at the International Broadcasting Convention (IBC) in 2018
and at the Mobile World Congress (MWC) in 2019. In order to demonstrate
feasibility of the developed solutions in operational conditions, 5G-Xcast also
aims to participate in a 5G large scale trial around the European Championships
in summer 2018. Furthermore, the project will demonstrate 5G-Xcast technolo-
gies for the PWS vertical4 in at least one of the urban city test-beds.
TIM, as an operator, is involved together with the other consortium partners in the
use case and scenario definition to be used as a reference for the overall project
investigations and experimentations. TIM is contributing to both the radio access
aspects and core network aspects. Concerning the Radio Access Network TIM’s
contribution is focused on radio electric coverage analysis and RRM. Regarding
5G XCAST 369
the Core Network TIM is involved in the investigations on architectural aspects.

Up to now, in general, mobile broadcasting has not been successful as expected.
In any case, in the light of the introduction of 5G that, through eMBB, enables the
provision of multimedia contents with a high quality level not possible with pre-
vious mobile systems generations, even the multicast/broadcast could gain a
higher interest. And, also for PWS (Public Warning Systems), an efficient mul-
ticast/broadcast 5G solution can represent an important way to provide such a kind
of services. The project results can then be very useful for TIM in order to better
exploit the cellular resources when designing and operating the future 5G net-
work. Also the new and innovative concepts studied in the project with a view to
achieving a coordinated usage of currently “isolated” networks to deliver mul-
ticast/broadcast services will provide TIM with guidelines for optimized network
design and operation based on usage scenarios and related requirements with pos-
itive impacts on both the customer experience and consequent business return.
5G-CROSSHAUL project
D. Siracusa, E. Salvadori, G. Landi, N. Ciulli, C. Casetti, A. Di Giglio, P. Io-

vanna12
Abstract The aim of 5G-CROSSHAUL project is to develop an adaptive, shara-

ble, cost-efficient 5G transport network solution integrating the fronthaul and
backhaul segments of the network whilst supporting existing and new radio access
protocol functional splits envisioned in 5G. This transport network will flexibly
interconnect distributed 5G radio access and core network functions, hosted on
in-network cloud nodes, through a unified data plane encompassing innovative
high-capacity transmission technologies and a control infrastructure facilitating
the set up of end-to-end services.
Domenico Siracusa, Elio Salvadori

FBK CREATE-NET
Giada Landi, Nicola Ciulli

Nextworks Srl
Claudio Casetti
Politecnico di Torino
Paola Iovanna
Ericsson Telecomunicazioni SpA
Andrea Di Giglio
TIM
372 D. Siracusa, E. Salvadori, G. Landi, N. Ciulli, C. Casetti, A.
Di Giglio, P. Iovanna
1 Project summary
5G-PPP Phase 1
H2020 Call H2020-ICT-2014-2
Project duration: 1 July 2015 – 30 December 2017
Project budget: € 8.352.273,00
Requested EC budget: € 7.942.523,00
Project coordinator: Universidad Carlos III (Spain)
The key idea of the 5G-Crosshaul proposal is to move from a traditional

transport network perspective to a novel architecture integrating both fronthaul
and backhaul into the 5G-Crosshaul transport network. Figure 1 shows the phys-
ical infrastructure that 5G-Crosshaul is composed of, categorized into three dif-
ferentiated layers. The bottom layer corresponds to the “Interconnection Plane”
and shows the networking infrastructure, formed by heterogeneous links connect-
ing the different elements of the planes located above.
The “Interconnection Plane” makes use of 5G-Crosshaul Packet Forward-
ing Elements (XFE) to interconnect a broad set of novel technologies to create
a packet-based network that can meet the demands of 5G networks. The tech-
nologies, which have already been identified as relevant for the future of the back-
hauling, are represented in the figure too. They span from fibre optics to novel
CPRI-over-packet technologies, also considering wireless links such as mmWave.
The second plane (depicted in green in Figure 1) has been named “5G-Crosshaul
General Processing Plane” and shows the different 5G-Crosshaul Processing
Units (XPUs) that carry out the bulk of the operations in the 5G-Crosshaul. The
functionality provided by these XPUs is multi-faceted. It highly depends on the
actual interconnection and must encompass functionality expected from the dif-
ferent elements in the 5G network (depicted in the uppermost layer of
Fig). As an example, in (partially) centralized RAN implementations, the XPUs
will host BBUs or MAC processors (thus enabling true Cloud RAN). Further, they
might serve as end-points for services or even caches. The different functional
distributions between 5GPoA and XPU relation and the different services that can
be hosted in the XPUs are one of the pillars of the flexibility provided by the 5G-
Crosshaul architecture.
5G-CROSSHAUL Project 373
Xhaul
End-Point Plane
Xhaul
General Processing Plane
Xhaul
Interconnection Plane
Xhaul Packet Forwarding Element

Fiber Optics Operator1 PoP
(XFE)
Wireless Sub-6GHz 5G PoA Operator2 PoP
mmWave eNodeB Legacy
microWave RRH
Copper Pico/Femto
Xhaul Processing Unit
Free Space Optics
(XPU)
Figure 1: Physical Infrastructure of 5G-Crosshaul
This is represented by the different connection options between the uppermost

(“End-Point Plane”) and the middle layer of Figure 1.
1 Project Objectives
The 5G-Crosshaul project is a very ambitious initiative aiming at designing the

transport network that will serve the 5G deployments. The next generation
transport network needs to unify the way it manages the different traffic sources,
with really diverse, and potentially extreme, requirements in terms of bandwidth,
latency or number of users. Specifically, the project pursues the following eight
key objectives:
• Design of the 5G-Crosshaul Control Infrastructure (XCI): Develop XCI by
extending existing Software Defined Network (SDN) controllers to provide
the services for novel Northbound (NBI) and Southbound (SBI) Interfaces
and enable multi-tenancy support in trusted environments.
• Specification of the XCI’s northbound (NBI) and southbound (SBI) inter-

faces: Define interfaces to accelerate the integration of new data plane tech-
nologies (SBI) and the introduction of new services (NBI) via novel or ex-
tended interfaces. In more details the work consists of:
o Specify an abstract network information model for 5G-Crosshaul tech-
nologies, including abstracted control parameters and system status met-
rics.
o Collect contextual information about traffic, congestion, interference of
radio resources, and energy savings to provide applications (WP4) with a
media-independent system status report of the heterogeneous network.
o Specify the set of southbound XCI actions (e.g., control the forwarding
behaviour, configure radio parameters, deploy/migrate 5G-Crosshaul
functions).
o Specify the set of northbound XCI actions (e.g., provisioning of new
VPNs) to enable Service Level Agreement (SLA)-level reports, create
new virtual 5G-Crosshaul slices for multi-tenancy support.
• Unification of the 5G-Crosshaul data plane: Develop a flexible frame format
to allow carrying fronthaul and backhaul on the same physical link to replace
different technologies with a uniform transport technology for both fronthaul
and backhaul.
• Development of physical and link-layer technologies to support 5G require-
ments: Exploit advanced physical layer technologies, not currently used in the
5G-Crosshaul network segment, as well as novel technologies, such as wire-
less optics, silicon photonics for miniROADM, flexi-PON, etc. to increase
coverage and aggregated capacity of integrated backhaul and fronthaul net-
works. Increase cost-effectiveness of transport technologies for ultra-dense
access networks.
• Design of scalable algorithms for efficient 5G-Crosshaul resource orchestra-
tion: Develop and evaluate management and control algorithms on top of the
XCI NBI that ensure top-notch service delivery and optimal 5G-Crosshaul
resource utilization.
• Design of essential 5G-Crosshaul-integrated (control/planning) applications:
Develop an ecosystem of most essential XCI NBI applications, both to sup-
port (prediction, planning, monitoring) and to exploit (media distribution, en-
ergy management) the 5G-Crosshaul resource orchestration functions.
5G-CROSSHAUL Project 375
• 5G-Crosshaul key concept validation and proof of concept: Demonstration

and validation of 5G-Crosshaul technology components which will be inte-
grated into a set of 5G testbeds in Madrid, Berlin, Barcelona and Taiwan.
2 Contributions from Italian partners
In this Section we provide a short overview of the major contributions from the
Italian partners to the project.
TIM has been leading the task on the architecture requirements and contrib-
uted to the definition of the stakeholders. It has also developed a dimensioning
tool and performed an analysis of cost and energy consumption in a cross-haul
metropolitan realistic network, demontrating the impressive Capex and Opex sav-
ings due to the novel Crosshaul architecture and the introduction of miniROADM
in the optical stratum.
Furthermore TIM has also validated some experiments set-up and provided a
transmission demo in a real link.
Ericsson Telecomunicazioni SpA has provided the system requirements and
the definition of use cases, mapping them onto transport requirements. On this
basis, it has realized an innovative C-RAN solution that guarantees low and de-
terministic latency at lower cost compared to the solutions used in current aggre-
gation and metro networks. The solution includes novel framing protocol, deter-
ministic circuit-based switching and a cost effective silicon photonics reconfigu-
rable optical add-drop multiplexer (miniROADM). To assess the solution, both
simulations and experimentation in a proof of concept demo have been provided,
integrating the system in the 5G-Crosshaul demonstrator in the 5Tonic testbed.
Ericsson Telecomunicazioni SpA has also acted as leader of WP2, regarding
the specification of the data-plane, and as Innovation Manager of the project, con-
tributing to an ETSI White paper and submitting two approved contributions at
ITU-T IMT-2020.
Networks has performed the design of the XCI components, with focus on the
internal components of the XCI SDN controller and the XCI NFV MANO stack.
In particular, Nextworks has implemented extensions to the OpenDaylight SDN
controller for the monitoring and control of the power-sates and the flow
switching with QoS in XPFE-based network domains. Moreover, the team has
also developed with Politecnico di Torino an energy-efficient algorithm for joint
resource orchestration of XPFEs and XPUs, and has evaluated its performance in
both emulated and real testbed (5TONIC in Madrid) environments.
Networks has integrated the SDN controller with the XPFEs network domain in
5TONIC testbed and implemented an NFV orchestrator and VNF Manager com-
pliant to the ETSI IFA standard specifications, which has been validated in the
lifecycle management of an OpenAirInterface vEPC network service. Public
demonstrations of the various developments on NFV MANO and integrated SDN
controllers have been executed in international conferences (EUCNC 2017) and
in the context of 5TONIC.
FBK CREATE-NET has contributed to the finalization of models for mm-
Wave radio and Ethernet links interconnecting 5G-Crosshaul switching elements
in a random topology network. It has also defined the REST APIs for service re-
quest, provisioning and flow actions of the Resource Management Application
(RMA) in the 5G-Crosshaul architecture. It performed the design and develop-
ment of the Resource Management application (RMA) and its integration with
other building blocks (i.e., TVBA, CDNMA and MMA applications) in one of the
final project demo.
Politecnico di Torino has developed experimental-based models of the energy
consumption due to different virtualization approaches and designed optimization
algorithms for on/off switching of network nodes, leading to energy saving. It has
also performed the design and development (on top of an ONOS-based SDN con-
troller) of the EMMA application, which efficiently routes traffic flows over the
network in order to minimize energy consumption.
References
1. D1.2 Final 5G-Crosshaul system design and economic analysis - http://5g-crosshaul.eu/wp-
content/uploads/2018/01/5G-CROSSHAUL_D1.2.pdf
2. D4.2 Final design of 5G-Crosshaul Applications and Algorithms - http://5g-crosshaul.eu/wp-
content/uploads/2018/01/5G-CROSSHAUL_D4.2.pdf
3. D7.3 Final Project Report - http://5g-crosshaul.eu/wp-content/uploads/2018/01/5G-CROSS-
HAUL_D7.3.pdf
COHERENT project
R. Riggio, E. Salvadori13
Abstract The exponential growth of mobile traffic, drastically increasing of net-

work complexity, and the strong need for inter-network coordination of wireless
network resources call for breakthroughs in control, coordination and flexible
spectrum management in 5G heterogeneous radio access networks. The COHER-
ENT project aims to address these problems by researching, developing and vali-
dating a novel control framework for future mobile networks. The key innovation
of COHERENT is to develop a unified programmable control framework to co-
ordinate the underlying heterogeneous mobile networks as a whole.
1 Project summary
5G-PPP Phase 1
H2020 Call H2020-ICT-2014-2
Project coordinator: VTT Technical Research Centre of Finland Ltd
Roberto Riggio, Elio Salvadori

FBK CREATE-NET
378 R. Riggio, E. Salvadori
2 Objectives
The innovative impact of the COHERENT project is in enabling a unified control

and coordination framework for heterogeneous mobile networks by combining
innovative approaches on abstraction of low layers in underlying mobile net-
works, software defined networking, and flexible spectrum management. The
COHERENT control framework has two unique features to deal with the insuffi-
ciency of current control solutions for inter-network coordination. First, theories
and methods to abstract the low layer network states and behaviors of different
underlying mobile networks are developed, which provides a simplified but suf-
ficient abstracted network view for network-wide control and resource coordina-
tion. Network abstraction will significantly reduce the signaling overhead, making
scalable network-wide control solutions feasible, and enable more flexible spec-
trum management, which are key for the success of 5G networks. Second, based
on the abstracted network view, common interfaces and software-development
kits will be developed to enable programmability in controlling and coordinating
heterogeneous mobile networks. The programmable control will provide opera-
tors a flexible and cost efficient way to implement new control functions and thus
to support new services. Specifically, the project pursues the following eight key
objectives:
• Lay theoretical foundation and develop practical implementation on the ab-
straction of physical and MAC layer states, behaviours and functions of vari-
ous radio access networks to enable simple, unified, scalable control and co-
ordination solutions for HMN (Heterogeneous Mobile Networks).
• Build flexible spectrum management upon the awareness of network condi-
tions through properly developed network abstraction. We aim to signifi-
cantly improve spectrum reuse in 5G networks by applying fine-grade, real-
time, adaptive and network-aware spectrum management in LSA and inter/in-
tra- operator spectrum sharing.
• Design and develop SDK based low-layer abstractions to enable programma-
ble control and coordination in HMN. The SDK will define as set of RAN
Application Programming Interface (API) and communication protocols, al-
lowing network application to configure and modify the behaviour of the un-
derlying RANs.
COHERENT Project 379
• Propose traffic steering and resource allocation algorithms based on the CO-
HERENT control framework to realize network-wide resource optimization,
and especially network energy saving by taking advantage of awareness of
network topology and states from the COHERENT control framework.
• Leverage physical and MAC layer abstraction to improve user cooperation
like D2D communications in HMN, develop related protocols and control
methods based on the proposed COHERENT control framework, and show-
case its ability to significantly improve D2D communications in HMN.
• Explore the low-layer abstraction and COHERENT control framework to sig-
nificantly improve mobility management and coverage extension in HMN
with particular attention to the moving relay/cell scenarios.
• Provide practical programmable solutions by carefully applying the CAP the-
orem in the COHERENT control framework. Leverage the SDK to enable
programmable control for radio resource provisioning in HMN.
• Investigate, design and implement different strategies to enable RAN sharing
between multiple virtual operators. Virtualization is an integral part of SDN
and must be taken into account at the early stages of abstractions design in
different settings including cellular, cloud and D2D. Performance and logic
isolation of concurrent slices of networking resources will be investigated and
adapted to COHERENT through the use of the SDK.
• Demonstrate the benefits of the COHERENT concept by validating the pro-
ject use cases in realistic conditions. Showcase the technical and business im-
pact of the project results using rich infrastructure and simulators/Emulators
platforms from partners’ and FIRE facilities. Provide solid standard contribu-
tion to 3GPP RAN1, RAN3 and ECC PT1.

FBK CREATE-NET has contributed to the project by providing support to its
5G-EmPOWER platform [4] which was used in one of the project demo platforms
implementing the COHERENT SDK. The demo showcased the first prototype of
the COHERENT integrated control plane for heterogeneous networks. It has also
380 R. Riggio, E. Salvadori
implemented probabilistic monitoring, channel busyness module, and channel

quality map for WiFi networks. FBK CREATE-NET has also extended its 5G-
EmPOWER by adding support to the LTE open source platform srsLTE. It has
also proposed new methods for spectrum sharing in the multi-tenant use case, and
provided system level simulation model to analyse multi-tenant RAN and spec-
trum sharing. As WP leader on the workpackage about the Programmable RAN,
FBK CREATE-NET contributed to several algorithms for end-to-end Wi-Fi RAN
slicing, for channel–aware user association in software–defined Wi–Fi networks,
for traffic–aware user association in heterogeneous RANs, and for flexible func-
tional splits in 5G network. It has also implemented a RAN slicing in WiFi and
LTE networks.
References
1. D2.4 Final release of the SDK - http://www.ict-coherent.eu/coherent/wp-content/up-
loads/2018/05/COHERENT_D2_4_v1.0.pdf
2. D5.2 Final specification and implementation of the algorithms for programmable radio access
networks - http://www.ict-coherent.eu/coherent/wp-content/uploads/2018/03/COHER-
ENT_D5.2_v1.0.pdf
3. D6.2 Final report on technical validation - http://www.ict-coherent.eu/coherent/wp-con-
tent/uploads/2018/08/COHERENT_D6_2_final.pdf
4. 5G-EmPOWER - http://5g-empower.io/
SESAME project
R. Riggio, E. Salvadori, D. Munaretto, A. Albanese
14
Abstract The project proposes the Cloud-Enabled Small Cell (CESC) concept, a
new multi-operator enabled Small Cell that integrates a virtualised execution plat-
form (i.e., the Light DC) for deploying Virtual Network Functions (NVFs), sup-
porting powerful self-x management and executing novel applications and ser-
vices inside the access network infrastructure. The Light DC will feature low-
power processors and hardware accelerators for time critical operations and will
build a high manageable clustered edge computing infrastructure. This approach
will allow new stakeholders to dynamically enter the value chain by acting as
‘host-neutral’ providers in high traffic areas where densification of multiple net-
works is not practical.
1 Project summary

FBK CREATE-NET
Daniele Munaretto
ATHONET
Antonino Albanese
Italtel
382 R. Riggio, E. Salvadori, D. Munaretto, A. Albanese
5G-PPP Phase 1
H2020 Call H2020-ICT-2014-2
Project coordinator: OTE AE (Greece)
2 Objectives
SESAME project targets innovations around three central elements in 5G: the
placement of network intelligence and applications in the network edge through
Network Functions Virtualisation (NFV) and Edge Cloud Computing; the sub-
stantial evolution of the Small Cell concept, already mainstream in 4G but ex-
pected to deliver its full potential in the challenging high dense 5G scenarios; and
the consolidation of multi-tenancy in communications infrastructures, allowing
several operators/service providers to engage in new sharing models of both ac-
cess capacity and edge computing capabilities. In order to “address” this need and
building upon the pillars of network functions virtualisation, mobile-edge compu-
ting (MEC) and cognitive management, SESAME’s main goal is the development
and demonstration of an innovative architecture, capable of providing Small Cell
coverage to multiple operators “as-a-Service”. SESAME envisages the logical
partitioning of the localised Small Cell network to multiple isolated slices as well
as their provision to several tenants. Moreover, in addition to virtualizing and par-
titioning Small Cell capacity, SESAME supports enhanced multi-tenant edge
cloud services by enriching Small Cells with micro-servers. Specifically, the pro-
ject pursues the following nine key objectives:
• Definition and specification of the system architecture and interfaces for the
provisioning of multi-operator Small Cell networks, optimised for the most
promising scenarios and use cases.
SESAME Project 383
• Specification, design and implementation of a multi-operator CESC proto-

type, supporting Self-x features enabling multi-tenant and multi-service ac-
cess infrastructure.
• Specification, design and implementation of a low-cost Light DC prototype
as NFV Point-of- Presence (PoP), providing support for intensive low latency
applications, secure connections and high quality of experience, while mini-
mising space, infrastructure costs and energy consumption.
• Launching of at least three high-impact sample VNFs for demonstration and
assessment of SESAME CESC platform – Small Cell virtualisation through
providing MOCN as a VNF.
• Design and development of a framework for efficient resource planning and
coherent management of the multi-operator Small Cells as light NFV distrib-
uted infrastructure.
• Design and implementation of CESCM, capable of chaining and orchestrating
the different VNFs required to cope with the dynamic SLAs between the
CESC provider and the network operators.
• Integration into one Pre-Commercial Prototype of Small Cells, Light DC and
management and orchestration functionalities - The prototype will be used to
perform testing and evaluation, against solidly defined use cases.
• Conduct of Market Analysis, Road-mapping, as well as establishment of new
Business Models - Detailed Techno-economic Analysis towards exploitation
and commercialisation by industrial partners is also a priority.
• Communication/dissemination of project results raising awareness and im-
pact on stakeholders and the wider community, as well as assessment of re-
ceived feedback - Creation and exploitation of synergies with the rest 5G-PPP
projects and the Association towards building a consistent 5G view.
In this Section, we provide a short overview of the major contributions from the
Italtel has contributed to the definition and specification of the SESAME over-
all system architecture, especially focusing on the HW and SW design of the Light
384 R. Riggio, E. Salvadori, D. Munaretto, A. Albanese
DC. As WP leader of the work package 4 “Light DC Design and Implementation”,

Italtel has supervised all the contributions from the partners. Furthermore, Italtel
has designed and developed the computing and networking extension related to
the micro server virtualization SW platform, the video Transcoding service VNF
on both Intel and Arm platforms (including support for HW acceleration, Service
Function Chain, Monitoring and Scaling). Lastly, Italtel has provided the testbed
for the integration of the Light DC prototype used for one of the SESAME final
demo (benefits and performances of GPU HW acceleration in the video trans-
coding VNF).
FBK CREATE-NET has contributed to the project by contributing to the over-
all SESAME architecture definition. It has also contributed to define and consol-
idate the small cell functions virtualised within the architecture as well as to
demonstrate several CESC functions within the SESAME platform (caching of
multimedia content within the CESC, functional split and RAN sharing). As a
contribution toward the virtualised architecture, FBK CREATE-NET has also de-
veloped a software agent that allows the ClickOS virtual machines to communi-
cate with its own 5G-EmPOWER Virtualised Infrastructure Manager (VIM) plat-
form [3]. It has also also consolidated the software-agent that enabled the com-
munication between the Open Air Interface open source software, which imple-
mented the protocol stack of the SESAME small cell, and the 5G-EmPOWER
VIM. Finally, FBK CREATE-NET has developed a theoretical work on optimal
dynamic placement of small cell functions for multi-tenancy trying to minimize
aggregate interference.
ATHONET has contributed to the design of the SESAME use cases, as well
as on the definition of the SESAME network architecture. It has also participated
to the design and implementation of virtualized small cell systems by studying
and identifying the edge modules and components for MOCN from the core net-
work point of view. It has also been evaluating the feasibility (3GPP compliancy)
to implement the demonstrator designed and built in the work package and review
the interfacing proposed with the core network. ATHONET has integrated multi-
ple vEPCs with partners’ small cells from IPA and SISTEL, remotely and locally
with different RAN sharing and MOCN configurations, to deploy and test the
MOCN feature which enabled the use cases envisioned in SESAME.
SESAME Project 385
References
1. D2.5 SESAME Final Integrated Architecture and PoC Assessment KPIs - http://www.ses-
ame-h2020-5g-ppp.eu/Portals/0/Deliverables/SESAME_Deliverable%202.5_v1.0_Fi-
nal.pdf?ver=2018-10-29-153359-757
2. D7.5 Overall Assessment and Roadmap - http://www.sesame-h2020-5g-ppp.eu/Portals/0/De-
liverables/SESAME_Deliverable%207.5_v1.0_Final.pdf?ver=2018-11-08-142833-237
5G-ESSENCE project
R. Riggio, E. Salvadori, D. Munaretto, A. Albanese, M.R. Spada 15
Abstract The project addresses the paradigms of Edge Cloud computing and
Small Cell as a Service by fuelling the drivers and removing the barriers in the
Small Cell market, forecasted to grow at an impressive pace up to 2020 and be-
yond and to play a key-role in the 5G ecosystem. 5G ESSENCE provides a highly
flexible and scalable platform, able to support new business models and revenue
streams by creating a neutral host market and reducing operational costs by
providing new opportunities for ownership, deployment, operation and amortiza-
tion. 5G ESSENCE leverages knowledge, SW modules and prototypes from var-
ious 5G-PPP Phase-1 projects, SESAME being particularly relevant. Building on
these foundations, very ambitious objectives are targeted, culminating with the
prototyping and demonstration of 5G ESSENCE system in three real-life use
cases associated to vertical industries.

FBK CREATE-NET
Daniele Munaretto
ATHONET
Antonino Albanese
Italtel
Maria Rita Spada

Wind Tre S.p.A.
388 R. Riggio, E. Salvadori, D. Munaretto, A. Albanese, M.R.
Spada
1 Project summary
5G-PPP Phase 2
H2020 Call H2020-ICT-2016-2
Project duration: 1 June 2017 – 30 November 2019
Project coordinator: OTE AE (Greece)
2 Objectives
The technical approach envisioned by 5G ESSENCE project exploits the benefits

of the centralisation of Small Cell functions as scale grows through an edge cloud
environment based on a two-tier architecture: a first distributed tier for providing
low latency services and a second centralised tier for providing high processing
power for computing-intensive network applications. This allows decoupling the
control and user planes of the Radio Access Network (RAN) and achieving the
benefits of Cloud- RAN without the enormous front-haul latency restrictions. The
use of end-to-end network slicing mechanisms will allow sharing the 5G ES-
SENCE infrastructure among multiple operators/vertical industries and customis-
ing its capabilities on a per-tenant basis. The versatility of the architecture is en-
hanced by high-performance virtualisation techniques for data isolation, latency
reduction and resource efficiency, and by orchestrating lightweight virtual re-
sources enabling efficient Virtualised Network Function placement and live mi-
gration. Specifically, the project pursues the following eight key objectives:
• To specify the critical architectural enhancements from 5G-PPP Phase-1 that
are needed to fully enable cloud-integrated multi-tenant small cell network-
ing.
5G-ESSENCE Project 389
• To define the baseline system architecture and interfaces for the provisioning
of a cloud-integrated multi- tenant small cell network and a programmable
RRM controller, both customisable on a per vertical basis.
• To develop the centralised SD-RAN controller that will program the radio
resources usage in a unified way for all CESCs (Cloud-Enabled Small Cells).
• To exploit high-performance and efficient virtualisation techniques for better
resource utilisation, higher throughput and less delay at Network Services cre-
ation time.
• To develop the orchestrator enhancements for distributed service manage-
ment. The 5G ESSENCE orchestrator will jointly consider the decentralised
data centre environment with the operation of cSD-RAN controller for mul-
titenancy.
• To demonstrate and evaluate the cloud-integrated multi-tenant small cell net-
work in three real-life use cases associated to vertical industries, i.e., edge
network acceleration in a crowded event, mission critical applications, and in-
flight communications.
• To conduct a market analysis and to establish new business models. Detailed
techno-economic analysis and roadmapping towards exploitation and com-
mercialisation by industrial partners is also a priority.
• To ensure maximisation of 5G ESSENCE impact to the realisation of the 5G
vision by establishing close liaison and synergies with 5G-PPP Phase-1 & 2
projects and the Association. To pursue extensive dissemination and commu-
nication activities, as well to assess the perceived impact from the stakehold-
ers and the wider community.
In this Section, we provide a short overview of the major contributions from the
Italtel is participating in the elaboration of Use Cases and Requirements as
well as in the definition of the overall architecture that will specify how the SES-
AME Light DC will evolve towards the two-tier “Edge DC”. It is also participat-
ing in the design, development and testbed implementation of the Network
390 R. Riggio, E. Salvadori, D. Munaretto, A. Albanese, M.R.
Spada
Embedded Edge Cloud that includes the Edge DC (Light DC plus Main DC).
Italtel is involved also in the development of efficient, high performance virtu-
alised edge nodes on single-board computers (SBCs) based on both x86 and ARM
processors, supporting ultra-low latency virtualization platform. Lastly, Italtel is
contributing in the project Use Cases integration, demonstration and validation,
especially for the In-Flight Entertainment and Connectivity (IFEC) one, providing
the “i-EVS” (Enhanced Video Service) virtualized application used for the wire-
less in-flight streaming.
FBK CREATE-NET is involved in the design of the 5G-ESSENCE overall
architecture, and is contributing to the technical specification of the system, in
particular for the 5G cSD-RAN controller. FBK CREATE-NET will develop a
CESC slicing mechanism based on ClickOS and Docker approaches, and algo-
rithms to service-oriented resource assignment to the slices (both network and
licensed/unlicensed spectrum) and for scheduling the slices over the common
CESC infrastructure relying on different functional splits between virtual and
physical small cell. FBK CREATE-NET will contribute with a test-bed including
programmable small cells based on Open Air Interface (OAI), Wi-Fi and the 5G-
EmPower RAN controller [4] to perform preliminary tests in lab testing of slicing
and SON features.
Athonet is contributing to the use case definition and requirements. Athonet
mainly provides the core network functionalities (vEPC, vMBMS) to enable
broadcast video services during a live event in a stadium. Moreover, Athonet con-
tributes to the project by providing MEC functionalities at the edge of the network,
which represent a first bridge to the 5G technology across the use cases of the
project. Athonet also contributes to the standardisation efforts with a particular
focus on ETSI MEC and ETSI MCPTT.
Wind Tre is contributing to the requirements specifications and the definition
of the 5G ESSENCE architecture and in the elaboration of the Use Cases. Wind
Tre is guaranteeing also a part of the evaluation of 5G ESSENCE project in the
telecommunications domain; all the activities are performed through an investi-
gation of the new evolutions proposals for all the technological and procedural
aspects and the new possible interaction models, implemented by the project. Fi-
nally, Wind Tre is leading the workpackage (WP8) on “Dissemination, Commu-
nication & Exploitation of Results”, and is participating in the formulation of pre-
liminary business cases. Together the other industrial partners, Wind Tre is in-
volved in getting the best out of the project’s outcomes.
5G-ESSENCE Project 391
References
1. D2.1 System Use Cases and Requirements - http://www.5g-essence-h2020.eu/Por-
tals/0/5G%20ESSENCE_%20Deliverable%202.1_v1.0_Final.pdf?ver=2018-10-04-143227-
963
2. D3.1 Techniques for advanced radio resource centralised management and 5G network coor-
dination - http://www.5g-essence-h2020.eu/Portals/0/5G%20ESSENCE_%20Delivera-
ble%203.1_v1.0_Final.pdf?ver=2018-10-04-144118-413
3. D4.1 Optimisation of virtualisation, orchestration, and resource allocation - http://www.5g-
essence-h2020.eu/Portals/0/5G%20ESSENCE_%20Deliverable%204.1_v1.0_Fi-
nal.pdf?ver=2018-10-04-144621-157
5G-CARMEN project
R. Riggio, E. Salvadori, C. Casetti, M. Dell’Amico, L. D’Orazio, L. Andreone,

M. Cecchi, L. Pulcinelli, I. De Biasi, G. F. Cordella, S. Capato, B. Ferraioli, C.
Pahl16
Abstract The project will build a 5G-enabled corridor from Bologna to Munich
to conduct cross-border trials of 5G technologies in three major use cases: vehicle
manoeuvre negotiation (at various levels of automation), infotainment, and emis-
sion control. The 5G New Radio will be used to support latency sensitive and/or
bandwidth hungry services and applications. The project will leverage on a dis-
tributed mobile edge cloud spanning from the vehicle itself to the centralised
cloud. Multi-tenancy and neutral host concepts will be leveraged upon to deliver
a final platform capable of enabling new business models. 5G-CARMEN will
complement C-V2X with LTE and C-ITS technologies, targeting interoperability
and harnessing a hybrid network.
1 Project summary
R. Riggio, E. Salvadori (FBK) I. De Biasi (AutoBrennero)

C. Casetti, M. dell’Amico (CNIT) G. F. Cordella (DriveSec)
L. D’Orazio, L. Andreone (CRF) S. Capato (SWARCO)
M. Cecchi (TIM) B. Ferraioli (PIIU)
L. Pulcinelli (INWIT) C. Pahl (UNIBZ)
394 R. Riggio, E. Salvadori, C. Casetti, M. Dell’Amico, L.
D’Orazio, L. Andreone, M. Cecchi, L. Pulcinelli, I. De
Biasi, G. F. Cordella, S. Capato, B. Ferraioli, C. Pahl
5G-PPP Phase 3
H2020 Call H2020-ICT-18-2018
Project duration: 1 November 2018 – 31 October 2021
Project coordinator: Fondazione Bruno Kessler (FBK)
2 Objectives
European mobility is drastically changing: growing urbanisation, environmental

aspects, and safety are only a few of the key indicators pointing in this direction.
Road infrastructures and vehicles are blending with the digital world, becoming
always-connected, automated and intelligent, delivering optimal experience to
passengers, and addressing societal goals (e.g., emission and accident reduction)
and economic needs (e.g., vehicles as smart-living environments). In this respect,
the European Union pushes for large-scale collaborative cross-border validation
activities on cooperative, connected and automated mobility.
5G-CARMEN addresses these challenges harnessing the concept of “Mobility
Corridors”. In 5G-CARMEN important European industries, academics and in-
novative SMEs commit to achieve world-wide impact by conducting extensive
trials across an important corridor (by people/goods traffic volumes), from Bolo-
gna to Munich, spanning 600 km of roads, connecting three European regions
(Bavaria, Tirol and Trentino/South-Tyrol) across three countries. 5G-CARMEN
will realise a 5G-enabled corridor to validate a set of innovative Cooperative,
Connected, and Automated Mobility (CCAM) use cases from both business and
technical perspectives. To achieve this, 5G-CARMEN will leverage on the most
recent 5G technology enablers, including 5G NR, C-V2X interfaces, Mobile Edge
Computing (MEC), end-to-end network slicing, highly accurate positioning and
timing, and predictive quality of service. The neutral host model will be used in
order to enable this vision. Mobile Virtual Network Operators (MVNOs), Over-
5G-CARMEN Project 395
the-Top (OTT) providers, and service providers will have access to a multi-tenant
platform that supports the automotive sector transformation towards delivering
safer, greener, and more intelligent transportation with the ultimate goal of ena-
bling self-driving cars.
Specifically, the project pursues the following seven key objectives:
• Specify the technical and business requirements for the planned use cases in
order to enable an extended 5G CCAM corridor between Bologna-Munich.
• Design a federated 5G system architecture and deployment model that can
effectively support the CCAM use cases.
• Advance the enabling technologies for a 5G corridor including the network-
embedded cloud, the NR, and the new frequency planning to conform with
the CCAM KPIs.
• Design, implement, and validate a service-oriented, federated, and secure
platform for CCAM.
• Validate, evaluate and demonstrate the 5G-CARMEN concepts via single-
country and cross border pilots.
• Conduct a market analysis and to devise new business models based on the
5G-CARMEN platform. To provide a detailed techno-economic analysis and
road-mapping towards exploitation and commercialisation by industrial part-
ners.
• Ensure the long-term success of the project through standardisation and dis-
semination in commercial, industrial and scientific fora and by exploiting syn-
ergies with other 5G-PPP initiatives and projects.
Fondazione Bruno Kessler will act as project manager of 5G-CARMEN and
will involve two of its research centers: CREATE-NET and ICT. FBK will con-
tribute to the definition of the requirements of 5G-CARMEN use-cases and to the
detailed specification of system level architecture. FBK will also develop MEC
applications with special emphasis on low resource utilization and lightweight
virtualization using containers and container orchestration solutions, and it will

work towards network driven integration of C-V2X and ITS-G5 technology com-
ponents. FBK will develop mobile identity management solutions striking a good
balance between security, privacy, and user experience while checking the com-
pliance of the security and privacy policies with existing legal provisions in the
EU about processing personal data. FBK will also investigate centralized, distrib-
uted and hybrid realizations for resource control and orchestration across the het-
erogeneous physical and virtual infrastructure and contribute to the development
of required multi-domain service orchestration solution. Finally, FBK will con-
tribute in the validation of test results by collecting the required data from the
pilots and validate them using statistical analysis and simulations.
FCA-CRF will contribute to the use case definition and use cases require-
ments analysis, and provide requirements for the integration of a 5G connected
system on-board the vehicle; it will also participate in the system level technical
specifications. FCA-CRF will develop the on-board applications to enable se-
lected 5G-CARMEN use cases on its vehicle prototypes, adapting in-vehicle func-
tionalities according to the 5G-CARMEN requirements. FCA-CRF will integrate
the on-board system and the developed applications with the in-vehicle network,
and set-up 2 vehicle prototypes for the demonstration of the 5G-CARMEN system
for connected vehicles and automated driving. It will also contribute to the testing
and validation methodology, planning and preparation. FCA-CRF will be in-
volved in exemption procedures and safety issues activities, mainly regarding the
regulation aspects and authorization procedures for automated driving, and will
also support partners for the 5G connectivity part. It will also be involved in IPR
and data management issues, considering the aspects related to in-vehicle data
when dealing with connected driving and vehicle-infrastructure communication.
FCA-CRF will finally contribute to the market analysis and business modelling
as OEM, and will evaluate possible exploitation opportunities of the 5G-CAR-
MEN results within FCA.
TIM will contribute to the definition and analysis of the 5G-CARMEN se-
lected use cases with the connectivity requirements analysis for the implementa-
tion in the automotive corridor. TIM will study the main technologies enabling
the deployment of the 5G-CARMEN use cases, by considering the evolution in
the standardization of the V2X systems and MEC-based vehicular communica-
tions system. TIM will investigate network options to allow interworking for ve-
hicles equipped with ITS-G5 and LTE-V2X. As the partner with direct
involvement in the Italian sector of the 5G-CARMEN corridor, TIM will operate
the mobile network infrastructure deployed along the motorway A22 and will con-
tribute in the definition and set-up of the experimental tests in the Italian segment
of the vehicular corridor selected by 5G-CARMEN. TIM will act as worpackage
leader of “Dissemination, Partnerships and Cooperation models, and standardiza-
tion” activities, ensuring the participation of the project to global leading events
in the 5G and automotive sector and a proper impact of 5G-CARMEN outcomes
towards the relevant SDOs (3GPP and ITU) and 5G-PPP groups (i.e. 5G Auto-
motive Working Group).
CNIT will contribute to the detailed definition of use-cases, focusing on as-
pects related to virtualization, slicing and orchestration, and interworking between
5G, ITS-G5, and LTE C-V2X technologies. From the analysis of those use cases,
it will contribute to identifying detailed requirements and their impact on the func-
tional and system architecture. CNIT will collaborate to the evaluation of the crit-
ical components in both C-V2X and ITS-G5-based network elements at the phys-
ical and access layers that guarantee low-latency communication, and will design
solutions capable of ensuring smooth migration from existing or upcoming tech-
nologies (ITS-G5, LTE, and LTE C-V2X) to vehicular 5G. CNIT will also work
on the definition of new slicing models toward the novel paradigm of “Vehicle as
Infrastructure” and contribute with a software stack for predictably exploiting
next-generation embedded computing devices to seamlessly integrate the 5G
communication stack with the autopilot software. CNIT will contribute to the de-
sign of the required multi-domain service orchestration solution, focusing on
cross-border interoperation of MEC entities. It will contribute to the planning ac-
tivity of test sessions for the different pilots with the aim of validating the KPIs
associated to different use cases. CNIT will support the design of the analytics
needed to compare the data communication and execution latency with bench-
marks collected. CNIT will support the development of a market analysis and the
development of novel business models proposing the canvas methods used in
other projects.
Autostrada del Brennero will be end-user of the project and will made the
infrastructure available for the deployment of the project, also contributing with
its experience in terms of large-scale demonstration and issues related to the im-
plementation of on-site tests. It will contribute in the definition of the use cases
and the analysis of their requirements and work on the integration of the notifica-
tions messages originating from cellular base stations into the C-ITS server
installed at the TCC, so as to enable it to broadcast the message also on the ITS-
G5 network.
INWIT will investigate a viable “neutral hosting” approach in collaboration
with MNOs which is applicable to both the traditional and new components of the
network architecture. INWIT will identify suitable sites for tests, define MNO
lease space requirements and possible ancillary service they might require (e.g.,
FO backhauling) coupling those requirements with inherent Service Level stand-
ards. It will also contribute to the business analysis and modelling and KPI defi-
nition respectively according to a "neutral hosting " approach.
SWARCO will provide solutions that enable the deployment of C-ITS ser-
vices, through the 5G technology. It will contribute to the definition of the use
cases requirements and specifications and of the functional architecture of the pro-
ject. It will also contribute to the integration of 5G technologies into the existing
solutions for connected vehicles, while providing 5G-CARMEN with its infra-
structure, products and knowledge to reach the successful integration and testing
at pilot level.
DSEC will provide to 5G-CARMEN the competence and resources needed to
provide to the solutions the needed and appropriate cybersecurity protection, con-
sidering regulation and functional safety. DSEC will provide guidelines to be con-
sidered in the development of future CCAM applications, in order to properly
address cybersecurity protection and a methodology to assess the level of cyber-
security, as a validation tool, of components or systems. DSEC will also partici-
pate to 5G-CARMEN pilots execution, to validate the methodology and to assess
the cybersecurity of the developed applications.
PIIU will elaborate specific licensing models regarding the part of the results
of the project that will be exploited under open sources schemes. It will also focus
on the social-economical analysis aspects of the 5G-CARMEN project. PIIU will
be the responsible for the creation and maintenance of the concrete plan for the
dissemination and use of knowledge of the project, by coordinating the dissemi-
nation actions of the project in scientific and business events and being responsi-
ble for the project’s visibility through online channels, such as the online social
media channels. PIIU will also manage the activities related to the involvement of
external SMEs to unlock the commercial potential of high-value solutions offered
by 5G-CARMEN.
UNIBZ will bring its expertise in the management of virtual resources in edge
computing environments. More specifically, the contribution is to address cross
domain orchestration, i.e. orchestration of resource between different clouds. The

5G-CARMEN project will involve clouds in different countries and owned by
different entities and workloads must be migrated from one cloud to the other
ensuring SLA aspects including security and performance related concerns.
References
1. 5G-CARMEN Internal documentation
2. 5G-PPP homepage: https://5g-ppp.eu/5g-carmen/
3. 5G-CARMEN official press-release on FBK website: https://magazine.fbk.eu/en/news/5g-car-
men-a-digital-corridor-for-the-mobility-of-the-future/
5GCity
Gino Carrozzo, Paolo Cruschelli, Maria Rita Spada, Mauro Di Bugno, Luca Vi-
gnaroli, Antonino Albanese, Andrea Michelozzi17
Abstract 5GCity is a H2020 Innovation Action funded by the European Commis-

sion as part of the 5G PPP Phase 2 programme. The project is designing, devel-
oping and deploying a distributed cloud and radio platform for municipalities and
infrastructure owners who can act as 5G Neutral Hosts. The main goal of 5GCity
is to build and deploy a common, multi-tenant, open platform that extends the
centralized cloud model to the extreme edge of the network, with a live demon-
stration and trials to be run in three different cities: Barcelona (ES), Bristol (UK)
and Lucca (IT). In these cities, three main categories of use case scenarios are
going to be tested and validated: Neutral Host, Media and Entertainment and City
Monitoring.
1 Project background and architecture

The value creation in the telecommunications market which is preparing to
5G roll-out is no longer the case of a closed market limited to a few players: new
roles and relationships are emerging to become essential part of the new 5G-
17
Gino Carrozzo and Paolo Cruschelli, Nextworks S.r.l.
Maria Rita Spada, Wind Tre
Mauro Di Bugno, Comune di Lucca,
Luca Vignaroli, RAI
Antonino Albanese, Italtel
Andrea Michelozzi, Comunicare Digitale
402 Gino Carrozzo, Paolo Cruschelli, Maria Rita Spada, Mauro
Di Bugno, Luca Vignaroli, Antonino Albanese, Andrea Mi-
chelozzi
powered ecosystem. As depicted in Fig, 1, the traditional roles of Hardware

(Equipment) manufacturers/vendors and Network Operators are being comple-
mented by Infrastructure owners, who own the necessary infrastructure that can
be used for hosting the computing, storage and networking infrastructure of the
5G network (e.g. space in street cabinets, lampposts or buildings along with power
supply and connectivity).
Fig. 1 Relationships between actors in the new 5G ecosys-

tem
A new role is taken by the Software Developers who develop the necessary soft-
ware as well as the Virtual Network Functions (VNFs) that are necessary to the
delivery of the 5G network services and to the specific requirements of the vertical
applications running on the 5G infrastructure (e.g. video analytics or augmented
reality software). Finally, Content/Service Providers are heavily entering the 5G
value creation chain, being them among the primary actors who realize the vertical
industry and who develop services and contents to be offered to End users, i.e.
the citizens.
5GCity 403
5GCity [1] has its main stakeholder in the municipalities who quite often own
and manage the best urban spaces to host 5G Small Cells and are undergoing a
digital transformation towards becoming Smart Cities. Here 5GCity generates
most of its im-
pact, unleashing
the power of new
value-added ser-
vices through the
support of a Neu-
tral Host model
(see
).
Fig. 2 5GCity Neutral Host concept
5GCity is designing a completely de-centralised, 3-tier architecture where

compute, storage and networking are allocated between core and edge segments
of the 5G network infrastructure capable to offer tight-loop interactions between
applications and enabling connectivity services [2][3]. The 5GCity functional ar-
chitecture is depicted in Fig. 2, which provides a view of the different functional
blocks necessary to adapt city infrastructure to the cloud and virtualized edge net-
work. This architecture is composed by three layers:
• Service/Application Layer
• Orchestration & Control layer
• Infrastructure layer
chelozzi
Fig. 2 5GCity Functional Architecture
The 5GCity architecture adapts distributed cloud technologies in order to

build its combined edge and network infrastructure providing a multi-tenant, cost-
effective platform for deploying virtualized heterogeneous services. Through this
architecture it is possible to realise the 5G Neutral Host model, i.e. the network
slicing mechanisms developed in 5GCity which allows to manage a network in-
frastructure to host any entity that uses it to provide its services to its end users.
In this model, the “Neutral Host” (infrastructure owner) is able to operate a parti-
tion of its resources and to arrange them in a set of homogeneous tenants (or
slices). Following this business model, future operators will interact with the fol-
lowing actors in order to monetize 5G investments and increase their revenues
from MNOs and vertical industries.
The 5GCITY platform is the core engine to implement the Neutral Host func-
tionalities. It is a software suite which allows complete lifecycle management of
end-to-end network services dynamically created over heterogeneous city infra-
structures. The 5GCITY platform [3] is composed of the following main elements:
• 5GCity /Orchestrator is the core entity of the platform, which handles
the complete life-cycle management of the network services, providing
the suitable degree of abstraction between resources as exposed by hard-
ware layer and services as intended by end-users.
5GCity 405
• 5GCity Dashboard is the main operational interface allowing 5GCity us-

ers (vertical or service providers) to deploy network services over on
owned set of network slices.
• 5GCity VIMs are the entities which acts as intermediary between the or-
chestrator and the physical resources located at various levels (Core VIM
for data center level, Edge VIM for network edge level, and Extended
Edge for devices level).
• 5GCity NFVIs are thin layers deployed over the pool of hardware re-
sources and acting as agent for the configuration of the physical resource
management (on the three levels, similarly to the VIMs).
• SDN controllers are the functional entities which provide suitable control
plane operations.
To ease the design and development of applications and services on this orches-
tration layer it is essential to
• Expose powerful APIs through which it is possible to access, define and
program the different edge services and the orchestrator functionalities.
• Grant access via APIs and via a service Software Development Kit to a
rich set of primitive functions for network and vertical application layer
services (e.g. programmable connectivity with QoS, media acquisition,
transcoding, traffic monitoring, security, etc.). These primitive functions
can be composed to form the various service flows the vertical applica-
tions requires.
Service development kits are a fundamental component to open-up the

virtualisation advantages to third party enterprises or individuals, also in terms of
commercial relationships. Different SDK toolkits are just available within the
NFV research (e.g. the COHERENT SDK, the SONATA SDK, etc.). However,
most of these SDKs adopt a network-centric approach, aiming at defining and
testing Network Services and VNFs before their instantiation in runtime MANO
environments. 5GCity advocates for a more data-centric programming model
based on Flow Based Programming (FBP), in which functional components can
be assembled into applications by connecting them using pipes. Under this model,
applications can adapt by rewiring components, adding/removing functions or
changing their implementation. 5GCity is extending the FBP model by using
smart-pipes, i.e. allowing in-network functions such as aggregation and filtering.
This can allow edge services to adapt to resource availability, handle failures, and
network configuration changes (for example, node migration).
chelozzi
In 5GCity, SDN/NFV network solutions, multi-access edge computing

(MEC), and optimized lightweight virtualization platforms are all integrated to
manage and orchestrate the distributed 5G platform deployed in the city. In fact,
other important features offered by the 5GCity architecture include:
• The ability to allocate or deallocate resources (Network slices and MEC
nodes) to the Service Providers according to their own SLA (QoS, etc.)
• The ability to dynamically reallocate resources to the Service Providers to
align with the changing needs of different users and applications
• The ability to reduce per-VM memory footprint (down to MBs or even
KBs);
• The ability to shorten creation times to support just-in-time deployment
of virtualized applications (e.g., in 100 of ms or less) and migration times
(down to 100 ms or less).
5GCity is working on these optimizations to allow to run a much larger number
of VMs than currently possible on resource constrained devices (e.g., in the order
of tens or more depending on the device). 5GCity will further provide a tool for
generating minimalistic, specialized VMs (a.k.a. unikernels) in order to provide
low overhead, virtualized edge applications in an automated way.
2 Pilots in 5GCities and use cases

5GCity is building city-wide pilots implementing three different types of use
cases, which simultaneously take place in dense environments such as the cities
and that need to co-exist and share the ICT infrastructure with different require-
ments: (i) neutral host; (ii) media industry use cases; and (iii) waste management
through video analytics [2].
The 5GCity Neutral host use case implements the category of telecom operator
driven use cases. It will leverage on orchestrated core/edge virtualization plat-
forms in order to enable the cities to create dynamic end-to-end slices containing
both virtualized edge and network resources and lease it to third-party operators.
For example, a 5G radio operator can become a service provider in cities where it
does not have infrastructure, with full management and control capabilities over
the slice, further than the typical MVNO use case.
5GCity 407
The second scenario (Industry vertical) is strictly related to different aspects of

the media and entertainment industry and encompass all the Use Cases pivoting
around video acquisition, editing and delivery. Three different media Use Cases
are taken into account in 5GCity:
• Mobile real-time transmission
• UHD video distribution
• Real-time video acquisition and production in edge cloud.
Instead the third scenario (City Services) is tailored to the specific needs about
security and surveillance of the territory in Smart Cities. 5GCity exploits the cit-
ies’ surveillance cameras and Video Analytics based applications that can process
video streams close to the acquisition cameras to rapidly identify illegal waste
dumping, violation of accesses, etc.
Fig. 3: 5GCity Use Cases Overview
Fig. 3 below depicts the logical perspective of the different use cases in 5GCity
which are being implemented and deployed in the cities of Barcelona, Bristol and
Lucca as pilot demonstrators. Six Use Cases have been identified corresponding
to the three vertical categories covered by 5GCity. Pilots for these use cases will
be developed in the three cities involved in the project as summarized in TABLE
I.
chelozzi
TABLE I. 5GCITY USE CASES AND SELECETED CITIES FOR PILOTS
City
ID Use Case Name Barce- Bris- Lucca
lona tol
UC1 Unauthorized Waste Dumping Prevention - - Yes
UC2 Neutral Host Yes Yes Yes
UC3 Video Acquisition and Production
Yes Yes -
Community media engagement in live events
UC4 UHD Video Distribution
- Yes Yes
Immersive Services
UC5 Mobile Backpack Unit for Real-time Transmis-
Yes - -
sion
UC6 Cooperative, Connected and Automated Mobil-
Yes - -
ity (CCAM)

In 5GCity, a solid group of Italian partners is leading the pilot validation ac-
tivities in Lucca and the supporting 5GCity platform development activities.
Nextworks is the Technical Manager of the project from Nov. 2018 and has
an active role in the 5GCity architecture design, in the specification of the various
NFV MANO interfaces, and is leading the activities related to the design and de-
velopment of the 5GCity SDK. Nextworks also participates in the configuration
of the 5GCity infrastructure in Lucca and in the development of the video surveil-
lance elements for the Unauthorized Waste Dumping Prevention Use Case.
Comune di Lucca, is the owner and maintainer of one of the three main pilot
sites of the project, directly involved in the activities of specification of the use
case and infrastructure requirements, classification and prioritization. Comune di
Lucca has the particular role of validator of the Unauthorized Waste Dumping
Prevention Use Case.
Italtel, coordinates the definition and specification of the 5GCity use cases
and architecture, as well as system high-level interfaces, workflows and business
5GCity 409
models. Moreover, Italtel contributes to the Virtualization and VNF data models
on MEC nodes, to the orchestration platform, including the related SDK, and to
the usage of machine learning techniques for improving service deployment.
Italtel is involved in the integration and validation activities of the 5GCity pilots,
particularly in Lucca. Moreover, Italtel is leader of the 5GCity Innovation Board.
RAI, as broadcaster, operator of terrestrial and satellite networks, and media
content provider in Italy, is developing the use case on UHD video distribution
and immersive services. RAI is using advanced video acquisition devices (e.g.
360 degree cameras) and a set of services for 4K/HDR video distribution to allow
the end-user to move in a city to obtain and immersive experiences in cultural
places like the City of Lucca.
Wind Tre, contributes to define 5GCity solutions requirements and use cases.
In Lucca Wind Tre is offering the spectrum for use case tests and contributes to
the verification, and testing of the 5GCity components over the city-wide infra-
structure.
Comunicare Digitale, contributes to the definition of the innovative UHD/4K
offering as digital services in the 5GCity pilot in Lucca, engaging 5G smart user
groups and evaluating results, reactions, and feedbacks. CODI is also the leader
of the 5GCity communication activities.
References
[1] H2020 5GCity Project web site, [Online]. Available: http://www.5gcity.eu/, Accessed on: Oct. 2018
[2] 5GCity Deliverable 2.1 - 5GCity System Requirements and Use Cases, December 2017
[3] 5GCity Deliverable 2.2 - 5GCity Architecture & Interfaces Definition, June 2018
5G-MEDIA: Programmable edge-to-cloud virtu-
alization fabric for the 5G Media industry
Francesca Moscatelli, Giacomo Bernini, Pasquale Andriani and Giuseppa Ca-

ruso18
Abstract The 5G PPP Phase 2 5G-MEDIA project innovates media applications

to exploit the capabilities of the underlying 5G networks and develops an inte-
grated SDN/NFV-based Service Virtualization Platform for the lifecycle manage-
ment and monitoring of media services in 5G networks.
1 Project objectives and architecture
Media applications are amongst the most de-manding services in terms of re-
sources, requiring huge quantities of network capacity for high bandwidth audio-
visual and other mobile sensory streams. 5G-MEDIA [1] is a 5G PPP project
which aims at innovating media-related applications by investigating how these
applications and the underlying 5G network should be coupled and interwork to
the benefit of both. The 5G-MEDIA framework leverages on SDN/NFV tools and
concepts for offering an integrated Service Virtualization Platform (SVP) capable
of handling the lifecycle management (LCM) and monitoring of media services
Francesca Moscatelli and Giacomo Bernini, Nextworks
Pasquale Andriani and Giuseppa Caruso, Engineering Ingegneria Informatica

412 Francesca Moscatelli, Giacomo Bernini, Pasquale Andriani
and Giuseppa Caruso
and applications in 5G networks. In particular, the project’s main goal is to deliver

an integrated programmable platform solution for the development, design and
operations of media applications in 5G networks by providing mechanisms to
flexibly adapt service operations to dynamic conditions and react upon events
(e.g. to transparently accommodate auto-scaling of services, VNF re-placement,
etc.). The 5G-MEDIA framework [2] depicted in Fig. 2 includes:
• an SDK tool for developing, validating and testing media-oriented ser-
vices,
• a generalized 5G Apps and Services Catalogue capable of validating and
storing heterogeneous 5G application ad service descriptors and translat-
ing them in NFVO-specific descriptors,
• an SVP including an ETSI MANO frame-work integrated with a Monitor
Adapt Plan and Execute (MAPE) component for an automated LCM of
media services based on collected and aggregated monitoring data,
• different Network Function Virtualization Infrastructures (NFVIs) com-
prising the “Physical Layer” that provide computing resources by differ-
ent operators and supporting different cloud technologies to allow the
instantiation of generic and media-specific VNFs, depicted at the “Vir-
tualized Resource Layer”, closer to the user (edge computing paradigm)
host.
Furthermore, the 5G-MEDIA platform enables the application of Function-
as-a-Service (FaaS) to VNF management, complementing traditional VM based
VNFs with FaaS based media specific functions, aiming at dramatically reducing
development cycles and slashing operational costs to 5G-MEDIA users. The com-
bination of the FaaS approach with the VNF packaging and the enablement of
inserting FaaS VNFs in a typical VNF forwarding graph is one of the main in-
novation aspects of the proposed 5G-MEDIA approach. Another specific innova-
tion of the 5G-MEDIA project is the integration of the 5G-MEDIA MAPE and
the related monitoring components able to gather NFVIs, VNFs and applications
metrics within the SVP. The Cognitive Network Optimiser (CNO) within the
MAPE is able to respond on dynamic changes of the environment (e.g., location
change of end users, varying QoS demands) and to adapt the deployment of VNF
forwarding graphs seamlessly to continuously meet expected QoS requirements.
5G-MEDIA: Programmable 413
edge-to-cloud virtualization
fabric for the 5G Media indus-
try
Fig. 2 5G-MEDIA Architecture

and Giuseppa Caruso
The 5G-MEDIA DevOps framework will be demonstrated through three me-

dia industry-oriented use cases [3].
Use case 1 - Immersive media and Virtual Reality
Fig. 3 5G-MEDIA Immersive media and Virtual Reality use case
Use Case 1 (UC1) is about the next generation of immersive media, and more
specifically, tele-immersive (TI) media where the focus is on capturing the actual
appearance and performances of users in real-time and transmitting it for remote
consumption. These services are the backbone of a variety of tele-immersive ser-
vices and applications like remote immersive exercising classes (e-health), aug-
mented reality (AR) inter-personal connections in life-size 3D representations
(communication), immersive emplacement of virtual spaces for sharing (social
virtual reality) or interaction (co-design).
The goal of the use case development activities in 5G-MEDIA are to assess the
feasibility of mechanisms to ensure the Quality of Experience of users in real-
time, interactive multi - party 3D immersive media applications.
try
Use case 2 - Smart Production and User‑generated Content
Fig. 4 5G-MEDIA Smart Production and User‑generated Content
Use Case 2 (UC2) is focused on evaluating how professional (remote) broadcast

productions can benefit from the advancement in 5G today and in the future. In
our scenario the broadcaster produces a live stream of the event and distribute
it (e.g. via its web channel): two cameras are capturing the event sending out
two typical broadcast signals in HD-SDI in the format 1080i50. To feed these sig-
nals into the network a converter/gateway (Physical Network Function) is used.
The gateway converts/compresses the signals from HD-SDI to IP. Then they are
fed into the 5G-MEDIA network (NFVI). Inside the network the media streams
are processed by different media-specific Virtual Network Functions (VNF) which
are instantiated upon request from the 5G-MEDIA catalogue and
launched/onboarded (automatically) in the nearest/most appropriate NFVI (5G-
MEDIA network) via the 5G-MEDIA MANO which gives directions to the NFVI
VIM.
The goal of the use case development activities is to provide broadcasters with
ad-hoc, scalable, flexible and time-saving production mechanisms for remote me-
dia content, and demonstrate reduction in costs, time and complexity for content
production, enabling exploitation of user generated media content.
and Giuseppa Caruso
_______________________________________________________
Use case 3 - Ultra High Definition over Content Delivery Network
Fig. 5 5G-MEDIA Ultra High Definition over Content Delivery Network
Use Case 3 (UC3) targets the UHD media delivery over virtualized content distri-
bution networks. In particular, we are developing a vCDN solution capable of
meeting the needs of the increasing media industry, where the distribution of dif-
ferent types of media contents, with a high volume of data depending on the media
quality, is increasing and involves an heterogenous set of devices (e.g. phones,
pads and TV screens) connected to the network. In order to guarantee a strong and
persistent quality of experience (QoE) across devices connected to distributed
vCDNs.
try
The final goal of the use case development activities is to deliver new capabilities
to distribute UHD content (4K and 8K) with an optimal consumption of resources,
with enhanced security, better experience for end users and new market opportu-
nities in content delivery.
_______________________________________________________
2 Contribution of Italian partners
Two Italian partners are involved in 5G-MEDIA: Engineering Ingegneria Infor-

matica and Nextworks.
Engineering - Engineering is the 5G-MEDIA project coordinator leading WP1

and will lead the definition of the 5G-MEDIA business models.
Nextworks - In 5G-MEDIA, Nextworks will work on micro-Cloud platforms and

the NFV MANO components for media service delivery, with focus on the Open
Repository and the generic Network Apps for 5G-MEDIA. Nextworks will also
lead the validation activities related to the High Demanding UHD over Open
CDNs Pilot.
References
[1] H2020 5G-MEDIA Project web site, [Online]. Available: http://www.5gmedia.eu/, Accessed on: Oct. 2018
and Giuseppa Caruso
[2]
5G-MEDIA Deliverable 2.3 - 5G-MEDIA Platform Architecture, August 2018
[3] 5G-MEDIA Deliverable 2.2 - 5G-MEDIA Requirements and Use Case Refinement, November 2017
blueSPACE
Giada Landi, Marco Capitani 19
Abstract The blueSPACE project is a H2020 5G-PPP Phase 2 project that aims
develop novel technologies and architectures for the 5G fronthaul network, able
to guarantee high capacity and flexibility in scalable infrastructures. blueSPACE
exploits the optical Space Division Multiplexing (SDM) technology, proposing
an SDM-based infrastructure with multi-core fiber, optical beamforming and an-
alogue Radio-over-Fiber fronthaul, orchestrated through an SDN/NFV platform
for service provisioning.
1 Project objectives, technical challenges and architec-

ture
The blueSPACE project (http://5g-transformer.eu/) aims to exploit photonics
technologies to address the increasing bandwidth demands expected in 5G fron-
thaul networks, while guaranteeing the flexibility, programmability and scalabil-
ity required by 5G communication networks. The final goal is to develop an effi-
cient 5G infrastructure supporting a 1000-fold increase in capacity compared to
4G, ultra-low latency and reliable connectivity for over 1 billion users and a
deeper network programmability for automated adaptation and resources sharing.
blueSPACE proposes a fronthaul solution that combines the adoption of Spa-
tial Division Multiplexing (SDM) and flexi-grid DWDM to exploit both spatial
and spectral resources, integrating Analog and Digital Radio-over-Fiber (A-RoF
Giada Landi, Marco Capitani

Nextworks
420 G. Landi, M. Capitani
and D-RoF) techniques. blueSPACE technologies can be integrated into optical

access networks for 5G infrastructures using massive beam steering capabilities.
On top of this novel fronthaul infrastructure, blueSPACE integrates an
SDN/NFV-based management and orchestration platform that adopts Multi-Ac-
cess Edge Computing (MEC) and Network Slicing concepts to deliver end-to-end
services across radio, fronthaul/backhaul and core network segments.
blueSPACE addresses several technical and research challenges, as follows:
• Adaptation of DRoF schemes and ARoF techniques into SDM networks,
with increased degree of integration and full compatibility between the
two schemes.
• Development of advanced hardware components for (i) cost efficient
ARoF transceivers; (ii) remote power distribution over SDM networks
for efficient Remote Radio Units (RRUs) powering; (iii) spatial optical
beam forming schemes, compatible with ARoF transceivers at the RRU;
and (iv) compact SDM splitters and Multi-Core Fiber (MCF) adapters for
advanced SDM-based Optical Distribution Network (ODN) designs.
• Implementation of algorithms and procedures for optimum allocation of
the Baseband Unit (BBU) pools and edge applications (e.g. video analyt-
ics, data caching) at the Central Office (CO), with joint configuration of
switching and interconnection in fronthaul and backhaul networks.
• Design of optimized interfacing elements between SDM technology and
radiating elements in the RRU sites to enable massive MIMO and beam
steering solutions for both DRoF and ARoF.
• Design and implementation of an SDN/NFV/MEC-based control and or-
chestration platform to support the deployment of virtual BBUs and edge
functions in the CO, with integrated configuration of the SDM/WDM-
enabled fronthaul and packet-based backhaul.
• Design and development of network slicing and multi-tenancy solutions,
integrated with the virtualization of fronthaul/backhaul network re-
sources, for deployment of services from different verticals over a shared
infrastructure.
The solution proposed by blueSPACE [1] for the fronthaul network architec-
ture is depicted in Fig 1. Multiple cell sites are connected to a CO through an
SDM/WDM-based ODN with ARoF and DRoF transceivers and Power over Fi-
ber (PoF) devices to power the cell sites. The CO includes a pool of BBUs or
Distributed Units (DU) connected to an array of ARoF and DRoF transceivers,
blueSPACE 421
and an optical switch connecting the ARoF/DRoF transceivers to an array of

Spectrum Selective Switches (SSS) for spectral multiplexing. Mixing SDM and
flexi-grid DWDM, this architecture exploits the spectral and spatial dimensions
in the fronthaul segment. The cell site deploys a spatial multiplexer/demultiplexer
with some ports connected to several ARoF/DRoF transceivers (spatial channels)
or one ARoF/DRoF transceiver (spectral channels).
At the CO, a small Data Center acting as an NFV infrastructure point of pres-
ence (NFVI-PoP) is adopted to deploy VNFs and edge applications (e.g. video
analytics, data caching, location services.), exploiting the concept of Multi-Ac-
cess Edge Computing. A packet network is adopted at the backhaul segment for
the connectivity among the BBU/DUs, the small DC, and the optical metro/core
network.
MCM 1
ARoF RRH
…
ARoF RRH
Edge Spatial N
Computing
channels
PoF Tx
Pool of
BBUs/DUs Optical Switch (OS)
BBU ARoF
1 SSS Cell site (b)
1
MCM SSS/AWG
BBU ARoF 1
M ARoF RRH
…
1
…
…
Optical 1
Spectral
…
SSS
Metro DU DRoF N channels
…
ARoF RRH
Network M
DU DRoF
M SDM/WDM PoF Tx
Packet ODN
Switch (PS)
PoF Tx
Cell site (c)

Pool of PoF Tx
RoF Tx MCM
…
1
…
DRoF RU
PoF Tx
Central Office
…
…
MCM Spatial or
Optical Switch (OS) Spectral DRoF RRU
channels + N
1
statistical DRoF RU
packet
…
multiplexing
DRoF RU
M
Packet
PoF Tx Switch (PS)
Fig 1. blueSPACE fronthaul network architecture
On top of this infrastructure, blueSPACE adopts an SDN/NFV platform which

extends the concept of the ETSI NFV Management and Orchestration (MANO)
framework to implement network service management and resource orchestration
functions. In particular, the mechanisms for resource orchestration are extended
422 G. Landi, M. Capitani
to support the integrated configuration of the 5G SDM/WDM-enabled fronthaul

through a dedicated SDN controller.

In blueSPACE, Nextworks (NXW) is leading the design and the development
of the NFV Platform and Network Slicing solution. In particular, Nextworks has
developed an NFV Orchestrator (NFVO) which integrates mechanisms for the
configuration of the transport network via SDN controllers and for the orchestra-
tion of MEC hosts’ resources to deploy MEC applications. In blueSPACE con-
text, this NFVO interacts with an SDN controller operating over the SDM/WDM-
based fronthaul and manages the MEC resources made available at the Central
Office. On top of the NFVO, Nextworks is also developing a tool for management
of network slices in a multi-tenant environment.
[1] I. Tafur Monroy et al., “Space Division Multiplexing 5G Fronthaul with

Analog and Digital Radio-over-Fiber and Optical Beamforming – the blueSPACE
Concept”, August 2018, available on-line:
https://static1.squarespace.com/static/594fd79fd482e9221ac80add/t/5b80165b0e
bbe8d5b4a00d96/1535120999031/Concept+paper+of+the+blueSPACE+pro-
ject.pdf
5G-CroCo
Gino Carrozzo, Giada Landi 20
Abstract The 5G-CroCo project is a H2020 5G-PPP Phase 3 Innovation Action

project, started in November 2018, that will trial 5G technologies in the cross-
border corridor connecting the cities of Metz-Merzig-Luxembourg, traversing the
borders between France, Germany and Luxembourg to validate advanced 5G fea-
tures such as 5G New Radio, MEC-enabled distributed computing, Predictive
QoS, Network Slicing, and improved positioning systems, all combined together,
to enable innovative use cases for connected, cooperative and autonomous mobil-
ity (CCAM).
1 Project objectives
5GCroCo is a new 5G PPP project started on Nov 1st 2018 to specifically trial
5G technologies in the cross-border corridor along France, Germany and Luxem-
bourg [1]. 5GCroCo also aims at defining new business models that can be built
on top of this unprecedented connectivity and service provisioning capacity. Ul-
timately, 5GCroCo will impact relevant standardization bodies from the telco and
automotive industries.
The possibility of providing connected, cooperative and autonomous mobility

(CCAM) services along different countries when vehicles traverse various na-
tional borders has a huge innovative business potential. However, the seamless
Gino Carrozzo, Giada Landi

Nextworks
424 G. Carrozzo, G. Landi
provision of connectivity and the uninterrupted delivery of services along borders

also poses interesting technical challenges. The situation is particularly challeng-
ing given the multi-country, multi-operator, multi-telco-vendor, and multi-vehi-
cle-OEM scenario of any cross-border layout. Motivated by this, 5GCroCo brings
together a strong consortium from both, European automotive and mobile com-
munications industries, with the explicit support of road traffic authorities and the
respective national governments (through letters of support), to develop innova-
tion at the intersection of these two industrial sectors. The aim is to define a suc-
cessful path towards the provision of CCAM services along cross-border scenar-
ios and reduce the uncertainties of a real 5G cross-border deployment. 5GCroCo
aims at trialling 5G technologies in the cross-border corridor connecting the cities
of Metz-Merzig-Luxembourg, traversing the borders between France, Germany
and Luxembourg.
5GCroCo is part of the pan-European network of 5G corridors now emerging

thanks to a number of regional agreements. Under these agreements, hundreds of
kilometres of European motorways have been allocated for automated driving
tests. They cover scenarios up to the third level of automation (SAE Automated
Driving Level 3), the stage where a car can operate itself, with a driver present
under certain conditions. Three use cases related to Cooperative Connected and
Autonomous Mobility (CCAM) services and enabled by 5G technologies will be
tested in the corridor, namely: 1) anticipated cooperative collision avoidance, 2)
generation and distribution of dynamic high-definition maps for autonomous driv-
ing, and 3) teleoperated driving.
These corridors are supported by the European Commission as part of its 5G
Action Plan [2]. This plan aims to ensure commercial deployment of 5G technol-
ogies by the end of this decade. As the first step, before large-scale deployment in
the corridor, 5GCroCo will deploy local pilot schemes in Munich city centre, on
the German A9 motorway and on the closed test track in Montlhéry, France, as
well as in Barcelona city centre, where a cross-border city scenario will be emu-
lated.
The objective of the project is to validate advanced 5G features, such as New
Radio, MEC-enabled distributed computing, Predictive QoS, Network Slicing,
and improved positioning systems, all combined together, to enable innovative
use cases for CCAM. 5GCroCo aims at defining new business models that can be
built on top of this unprecedented connectivity and service provisioning capacity,
5G-CroCo 425
also ensuring that relevant standardization bodies from the two involved indus-
tries are impacted.

In 5GCroco, Nextworks is working on core services for Network Slicing,
multi-domain resource orchestration, configurable monitoring and edge compu-
ting orchestration, adaptations/developments on MEC related network services
for tele-driving.
References
[4] H2020 5Groco Project web site, [Online]. Available: https://5gcroco.eu/, Accessed on: Dec. 2018
[5] European Commission, 5G Action Plan, [Online]. Available:https://ec.europa.eu/digital-single-mar-
ket/en/5g-europe-action-plan, Accessed on: Dec. 2018
SELFNET
Giacomo Bernini, Pietro G. Giardina
Abstract SELFNET proposes an autonomic network management framework to

meet the requirements of advanced self-organizing network (SON) capabilities
for the 5G network infrastructure. In particular, a combined architecture that inte-
grates SDN and NFV and network intelligence is provided.

SELFNET [1] provides an autonomic management framework based mainly
on the combination of Software-Defined Networking (SDN), Network Function
Virtualization (NFV) and network intelligence to provide self-organizing en-
hanced capacities to 5G networks. Following a use case driven design approach,
SELFNET provides smart autonomic management of NFV and SDN functions in
order to automatically resolve or mitigate network problems and improve the
QoS/QoE.
The development of the SELFNET framework has been driven by the speci-
fication of three independent, ambitious use cases that have played crucial roles
in realizing the highly autonomic network management capabilities required by
5G networks:
• Self-Protection Use Case: to detect and mitigate effects of cyber-attacks
and restore 5G network traffic to steady state of security by means of
dedicated SDN and VNF sensors and actuators (e.g. virtual DPIs, virtual
honeynets, virtual IPSs) deployed and chained at different locations of the
network
428 Giacolo Bernini, Pietro G. Giardina
• Self-Healing Use Case: to detect and predict common failures and mal-
functioning in 5G network infrastructures (at both hardware and software
level, including infrastructure and operational issues) to apply reactive or
proactive recovery measures by means of dedicated actuations
• Self-Optmization Use Case: to deliver high-quality video streams from a
network media server to end users in a hotspot covered by a 5G network
with guaranteed QoE, by detecting and mitigating network congestion ef-
fects through dedicated QoE video sensors and actuators
The SELFNET framework architecture [2] enhances the ETSI NFV architec-
ture approach and principles towards enabling self-organizing capabilities in 5G
networks management, and is built around five differentiated layers with the fol-
lowing logical scopes (as shown in Figure 1):
• Infrastructure Layer: This layer provides the resources required for the
instantiation of virtual functions (Compute, Network and Storage) and
supports the mechanisms for that instantiation. It represents the NFVI
(Network Functions Virtualization Infrastructure) as defined by the ETSI
NFV terminology.
• Virtualized Network Layer: This layer represents the instantiation of the
Virtual Networking Infrastructures created by the users of the infrastruc-
ture as part of their normal operational plan and those created by the
SELFNET framework as part of the SON capabilities. The layer is com-
posed by a number of network functions and applications interconnected
in a designed topology in order to provide the functionalities required by
the user
• SON Control Layer: This layer contains the applications that enables the
collection of data from sensors deployed through the entire system (SON
Sensors) and the applications that re responsible for enforcing actions into
the network (SON Actuators) as part of the enabling mechanisms to pro-
vide network intelligence in 5G networks.
• SON Autonomic Layer: This layer provides the mechanisms to provide
network intelligence. The layer collects from the network pertinent infor-
mation about the network behaviour, uses that information to diagnose
the network condition, and decides what must be done to accomplish the
system goals. It then guarantees the organized enforcement of the actions
SELFNET 429
that are determined. A Service Orchestrator plays a key role to glue the
different monitoring, analysis, decision and action enforcing functions to-
gether and close the SON control loop. Specific Onboarding features are
also provided to equip the SELFNET framework with SON sensors and
actuators network functions and applications.
• NFV Orchestration & Management Layer: This layer corresponds to the
ETSI NFV Management and Orchestration layer, and provides its basic
functions: NFV Orchestration, VNF Management, and Virtualised Infra-
structure Management. The aim is to manage the lifecycle of SON sensors
and actuators in a common and unified way.
In terms of applications, SELFNET accommodates a heterogeneous set of

sensors and actuators network functions in support of the envisioned use cases
(self-protection, self-healing, self-optimization). Several types of applications are
supported in SELFNET, all grouped under the concept of “APP”, which is the
SELFNET terminology used to represent all the network functions that can be
onboarded to the SELFNET platform [3]. An APP can be a Virtualized Network
Function (VNF, or more generally NFV application) or an SDN application, ac-
cording to the following definition:
• NFV applications: they implement in software either control or data plane
network functions on top of a virtualized environment. Therefore, they
are control plane VNFs or data plane VNFs managed by a cloud manage-
ment system, like OpenStack;
• SDN applications: they are software applications implementing control
plane functions that are not virtualized network functions, and can even
possibly run directly on top of a physical infrastructure. When these ap-
plications are deployed within the same runtime environment of an SDN
controller, they are referred as SDN Controller Apps; on the other hand,
when they are deployed and instantiated outside the runtime environment
of an SDN controller they are referred as SDN-Apps.
430 Giacolo Bernini, Pietro G. Giardina
Figure 6 SELFNET reference architecture

SELFNET 431
One Italian SME was involved in SELFNET, Nextworks.
Nextworks actively contributed to the definition and implementation of the self-

protection use case, with main focus on development of the related VNFs and
SDN applications. In particular, Nextworks provided the full NFV packaging of
a virtual IDS based on Snort, and a virtual Honeynet function including embedded
lifecycle management agents for Day1 and Day2 configuration of the VNFs.
Moreover, Nextworks developed an SDN application for selective network flow
control to be used in the context of the self-protection use case to support selective
mirroring, diversion and drop features. In addition, Nextworks developed the
Selfnet App Catalgoue, that maps to the onboarding component in the figure and
provide a full catalogue service for one-click and automated onboarding of VNFs,
SDN applications and PNFs in the Selfnet platform, with a common model and
approach following the ETSI NFV principles.
References
[1] SELFNET, https://selfnet-5g.eu/
[2] SELFNET D2.1, “Use Cases Definition and Requirements of the Systems
and its components”, https://doi.org/10.18153/SLF-671672-D2_1
[3] SELFNET D3.1, “Report and Prototype Implementation of the NFV &
SDN Repository”, https://doi.org/10.18153/SLF-671672-D3_1
SliceNet
Giacomo Bernini, Pietro G. Giardina, Ciriaco Angelo, Cosimo Zotti 21
Abstract SliceNet implements a modular framework for the management and

control of network slices in multi-domain 5G infrastructures, allowing verticals
to easily customize, configure and monitor their slice services.
SliceNet [4] aims at defining a vertical-oriented, Quality of Experience (QoE)-

driven network slicing framework that focuses on cognitive network management
and control of end-to-end network slicing operation and slice-based (and slice-
enabled) services in 5G networks. An effective and efficient orchestration, man-
agement and control of 5G network slices is crucial for delivering a wide plethora
of services customized for heterogeneous vertical businesses with often diverging
requirements for Quality of Service (QoS), QoE and Service Level Agreement
(SLA) at large. As a further challenge, SliceNet also tackle network slicing in
multi-domain and cross-provider environments, therefore requiring additional and
dedicated composition logics (at orchestration, management and control levels)
for slice providers to offer and deliver to verticals end-to-end slices while abstract-
ing the details of different providers involved.
Giacomo Bernini, Pietro G. Giardina - Nextworks

Ciriaco Angelo, Cosimo Zotti - Ericsson Telecomunicazioni
Giacomo Bernini, Pietro G. Giardina,
Ciriaco Angelo, Cosimo Zotti
While network slicing is being recognised as a fundamental technology in the

emerging softwarised and virtualised 5G mobile networks to enable independent
end-to-end logical networks on a shared physical infrastructure, still verticals and
slice consumers role and involvement in the runtime control and management of
their slices is very limited. In this context, SliceNet aims at going beyond this
current practice by targeting a high involvement of verticals and slice consumers
in all the phases of the end-to-end slice lifecycle, from slice/service definition, to
provisioning and runtime control and management. The main SliceNet idea refers
to the “Vertical-in-the-Loop” concept, that can be considered as one of the main
innovative principles proposed. With SliceNet, verticals and slice consumers are
able to have access to their slices and have the possibility to play on top of them
through a set of tailored control operations [5].
SliceNet lays its foundation on a set of key innovations: Network Slicing, One-
Stop API, Plug & Play control, Cognition and Cross-Plane Orchestration. The
overall architecture is highly based on based on Software-Defined Networking
(SDN) and Network Function Virtualization (NFV) technologies, and it is built
around three main logical planes (as depicted in Figure 1):
• Management Plane: encompasses all the mechanisms related with the de-
sign, deployment, provisioning, configuration, supervision and decom-
missioning of network resources, slices and services;
• Control Plane: includes the mechanisms required to guarantee that the
configurations applied by the management plane are respected and exe-
cuted during the slice and service delivery/run-time;
• Data Plane: responsible for processing and forwarding packets between
network elements.
One of the most important concepts of SliceNet is the capability to apply cognitive
management through enhancing an autonomous, policy-based, closed loop with
machine learning capabilities Errore. L'origine riferimento non è stata tro-
vata.. Two main policy-based autonomous closed loops are foreseen within the
SliceNet system architecture:
SliceNet 435
• Cognition Closed Loop & Autonomous Management Closed Loop:

a) this policy-based closed loop involves the Monitoring, Cognition
and the Orchestration Sub-Planes and is used for longer term pro-
active management procedures, such as QoE/performance sce-
narios;
b) besides implementing the autonomous management closed loop
described in a), this loop also includes the intelligent, e.g. ma-
chine-learning, procedures which, as a result, can update or create
new policies for both the autonomous management closed loop
(described in a)) and for the autonomous control closed-loop;
• Autonomous Control Closed Loop:
o this policy-based closed loop involves the Monitoring, Control
and Orchestration Sub-Planes and is used mostly for real-time
sensing and actuation on the network. It is based on policies and
models fed by the Cognition Sub-Plane.
SliceNet will demonstrate its verticals-oriented, QoE-driven network slicing

framework through three vertical use cases [7]:
• SmartGrid: it implements an advanced self-healing solution for electric

power grids for protection coordination, automatic reconfiguration and
differential protection;
• SmartCity: it implements a smart lighting service in a Smart City, over a
5G infrastructure, with slicing support, from IoT devices to the smart
lighting cloud application
• eHealth: where a connected ambulance acts as a connection hub for the
emergency medical equipment and wearables, enabling storing and real-
time streaming of video and scanned data to the awaiting diagnostic team
at the destination hospital.
Giacomo Bernini, Pietro G. Giardina,
Ciriaco Angelo, Cosimo Zotti
Figure 1 SliceNet logical architecture
Two Italian partners are involved in SliceNet: Ericsson Telecomunicazioni and

Nextworks.
Ericsson Telecomunicazioni as part of Ericsson Company, world leader providing

communications technology and services to enable 5G Networks, is contributing
to SliceNet with its strong background on SDN, NFV and telecommunication net-
works. Focus is on overall system architecture with specific effort on design, im-
plementation and integration of the: i) control mechanisms of 5G Radio and Core
Networks in order to achieve dynamic network slicing configuration in a virtu-
alised multi-tenant mobile edge computing environment, ii) functionalities and
modules to constantly monitor failures and to perform automatic re-configuration
SliceNet 437
of the resources for each slice. Microservices and Containers technology are used
for all produced SW modules.
Nextworks is bringing in SliceNet its long-term experience and proven back-

ground on SDN and NFV, focusing its contribution towards all the aspects related
to network slice orchestration and vertical slices customisation. In practical terms,
Nextworks is developing the Plug & Play control framework, following a micro-
service and containerized applications approach implemented with Kubernetes, to
offer vertical-tailored slice control instances to verticals, as isolated control envi-
ronments where customized control functions and logics are plugged to build a
truly tailored slice view with tailored runtime control features. Nextworks is also
actively contributing in the design and implementation of the Service and Slice
orchestration tools for combined coordination of end-to-end network slices and
NFV Network Services in support of the SmartCity, SmartGrid and eHealth ver-
tical use cases.
References
[4] SliceNet, https://slicenet.eu/
[5] SliceNet D4.1, “Plug & Play Control Plane for Sliced Networks”
[6] SliceNet D2.2, “Overall Architecture and Interface Definition”,
https://doi.org/10.18153/SLIC-761913-D2_2
[7] SliceNet D2.1, “Report on Vertical Sector Requirements Analysis and
Use Case Definition”, https://doi.org/10.18153/SLIC-761913-D2_1
5G-TRANSFORMER
Luca Valcarenghi, Barbara Martini, Claudio Casetti, Carla Fabiana Chiasserini,

Gino Carrozzo, Giada Landi, Marina Giordanino, Aleksandra Stojanovic, Paola
Iovanna, Giulio Bottari 22
Abstract The 5G-TRANSFORMER project is a H2020 5G-PPP Phase 2 project

that aims to facilitate verticals industries in provisioning and management of ver-
tical services through an SDN/NFV-based platform for the delivery of vertical-
tailored network slices over mobile transport networks.
Luca Valcarenghi, Barbara Martini

Scuola Superiore Sant’Anna
Claudio Casetti, Carla Fabiana Chiasserini

Politecnico di Torino
Gino Carrozzo, Giada Landi

Nextworks
Marina Giordanino, Aleksandra Stojanovic

Centro Ricerche Fiat
Paola Iovanna, Giulio Bottari

Ericsson Telecomunicazioni
440 L. Valcarenghi et al.
The 5G-TRANSFORMER project (http://5g-transformer.eu/) aims to trans-

form today’s mobile transport network into an SDN/NFV-based Mobile
Transport and Computing Platform (MTP), which brings the “Network Slicing”
paradigm into mobile transport networks by provisioning and managing
MTP slices tailored to the specific needs of vertical industries. The 5G-TRANS-
FORMER vision is built around two main objectives:
• Enabling vertical industries to easily meet their service requirements
through customized 5G end-to-end slices; and
• Aggregating and federating transport networking and computing fabric
resources, from the edge all the way to the core and cloud, to create and
manage MTP slices throughout a federated and virtualized infrastructure.
The 5G-TRANSFORMER solution [1] is based on extensions of the ETSI
MANO architecture and it consists of three novel building blocks depicted in Fig
1, namely a Vertical Slicer (VS) for service and slice management, a Service Or-
chestrator (SO) for service and resource orchestration in multi-domain, federated
scenarios and a Mobile Transport and Computing Platform (MTP) for abstraction,
virtualization and efficient allocation of 5G infrastructure resources.
Fig 1. 5G-TRANSFORMER architecture

5G-TRANSFORMER 441
The VS [3] provides a common entry point for all verticals into the system,
facilitating the specification, instantiation, monitoring and management of verti-
cal services. The VS dynamically creates and maps the services onto network
slices according to the verticals’ requirements, managing their lifecycle. It also
translates the vertical service and slicing requests into NFV network services
(NFV-NS) that are requested to the Service Orchestrator. In this sense, a slice will
be deployed as a NFV-NS instance.
The SO [4] implements service or resource orchestration and federation, de-
pending on the requests coming from the VS. Orchestration entails managing end-
to-end network services or resources that can be split into multiple segments de-
ployed in different administrative domains, based on service requirements and
resource availability. Federation entails managing administrative relations at the
interface between SOs belonging to different domains and handling abstraction of
services and resources.
The MTP [2], as the underlying unified transport stratum, is responsible for
providing the networking and computing resources required by the NFV-NS or-
chestrated by the SO. This includes their instantiation over the underlying physi-
cal transport network, computing, and storage infrastructure. The MTP may also
abstract the available resources, providing the SO with unified and technology-
independent views of the network topology.
2 Project impact
The project will demonstrate its outcomes applying its concepts and solutions
in several vertical industry use cases, as follows:
• Automotive: Autonomous Cruise Control (ACC) enforcement applica-
tion, Collaborative Advanced Driver Assistance Systems (ADAS) appli-
cations and Remote Vehicle Interaction (RVI) applications.
• eHealth: Improvement of the municipal emergency communication net-
work and development of a new technological solution for health work-
ers and volunteers.
• Media and Entertainment: media applications for stadia and major sport
events.
5G-TRANSFORMER targets several 5GPPP KPIs, such as reduced service

deployment times or OPEX and CAPEX reduction. The 5G-TRANSFORMER
project is an ambitious initiative that is planning to have strong impact in industry.
In fact, the project implements a plan for communication, dissemination, and ex-
ploitation to maximize its impact, which includes products and standardization
(e.g., ETSI, IETF and ONF). Therefore, innovation management is also a key
component. The technologies developed in 5G-TRANSFORMER are expected
to strengthen the position of European companies in the upcoming 5G Mobile
Network market, both in Europe and Worldwide, for the whole value chain
(Verticals, Operators, Service Providers, Manufacturers, SMEs, and Complemen-
tary Industries).

Several Italian partners are involved in 5G-TRANSFORMER: Centro Ricer-
che Fiat (CRF), Ericsson Telecomunicazioni (TEI). Nextworks (NXW), Politec-
nico di Torino, Scuola Superiore Sant’Anna.
Centro Ricerche Fiat (CRF) is one of the vertical industries participating to

the project. CRF is providing the use cases in the automotive sector, and their
detailed analysis. It is also developing, jointly with POLITO, EURECOM, SSSA,
and Ericsson, the automotive Proof of Concept (PoC) focusing on a service that
extends vehicle sensing capability in different road scenario such as at intersection
in the urban environment and developing some specific Virtual Network Func-
tions (VNFs). CRF will also provide real vehicles that will act as users of the PoC
and the real-world settings for the testing of the PoC and the measurements of key
performance indicators(KPIs) such as latency and reliability level of automotive
services.
Ericsson Telecomunicazioni (TEI) is leader of the workpackage (WP) dedi-

cated to the research and development of the MTP. The MTP is based on a single
logical point of contact abstracting and orchestrating computational, storage, and
network resources from the underlying Network Function Virtualisation Infra-
structure (NFVI) from different technological domains. The MTP interacts by
means of interfaces based in standard interfaces based on ETSI with both the
5G-TRANSFORMER 443
Service Orchestrator and the underlying Virtual Infrastrcuture Managers (VIM)

and Wide-Area Infrastructure Managers (WIM). TEI is also innovation manager
of the project.
Nextworks (NXW) is leading the design and the development of the 5G-
TRANSFORMER Vertical Slicer. The VS prototype is released as an Open
Source software under the Apache License v2.0. It offers a web-based GUI and
RESTful APIs to allow the verticals to select services from a wide catalogue of
blueprints offered by 5G-TRANSFORMER Service Providers, customize service
descriptors with vertical-driven requirements and QoS parameters, instantiate and
monitor the services at runtime. Internally, the VS implements procedures to
translates service instances into network slices that are jointly managed and coor-
dinated, arbitrating service concurrency based on verticals’ business requirements
and SLAs. At the southbound, the VS offers a plugin-based interface towards
multiple NFV Orchestrators, to integrate mechanisms for NFV network service
management. In 5G-TRANSFORMER, NXW is also responsible for the design
and development of a cross-layer Monitoring Platform, which enables the collec-
tion and elaboration of monitoring data about physical and virtual infrastructures,
network slices and services as enabler for service automation decisions. NXW is
also involved in 5G-TRANSFORMER PoCs in media/entertainment and automo-
tive sectors.
Politecnico di Torino (POLITO) is contributing to the development of all the

three 5G-TRANSFORMER architectural elements. In the VS POLITO is contrib-
uting in designing and implementing arbitration schemes between vertical service
slices. In the SO and MTP POLITO is contributing in designing and implementing
service placement algorithms considering both service requirements and different
level of resource abstraction. Jointly with CRF, EURECOM, and Ericsson,
POLITO is also developing a PoC for the implementation of an automotive slice
for safety services with URLL requirements, within the Multi-access Edge Com-
puting (MEC) environment. The PoC is based on the OpenAirInterface (OAI)
RAN and EPC and features several VNFs implemented through Virtual Machine
(VMs). It is also apt to combining real vehicles and emulated vehicles, with the
dynamics of latter ones being derived from simulations carried out using realistic
vehicular mobility.
Scuola Superiore Sant’Anna (SSSA) is contributing to the design and de-

velopment of both the SO and the MTP. In the SO SSSA is contributing to the
design and development of service orchestration algorithms. In the MTP SSSA is
developing advanced resilience schemes and interfaces between different techno-
logical domain controllers (e.g., ONOS, Openstack) for resource abstraction
based on standard ETSI interfaces, such as IFA 005. SSSA/CNIT Advanced Re-
search on Networking (ARNO) testbed is one of the four test sites of the project.
ARNO testbed features an SDN-controlled Data Center testbed, an SDN-con-
trolled wired/wireless access testbed, and an SDN-controlled IP/MPLS over Op-
tical Network testbed. In particular, ARNO features an installation of OpenAir-
Interface (OAI) for RAN and EPC with some Universal Software Radio Periph-
erals (USRPs). By accessing the ARNO testbed experimenters can combine
testbed components in a different and flexible way to reproduce complex network-
ing scenarios, spanning from Data Center and access network to aggregation/edge
network and optical core networks and different eNB functional splits.
All the Italian partners are involved in the Automotive proof of concept that
is one of the use case through which the project will show the effectiveness of the
proposed solution. In the Automotive proof of concept a service that extends ve-
hicle sensing capability at intersection (extended ICA) will be deployed by ex-
ploiting the 5GT framework.
[1] X. Li et al., “5G-TRANSFORMER Initial System Design”, 5G-TRANS-

FORMER Deliverable D1.2, May 2018
[2] C. Turyagyenda et al., “Definition of the Mobile Transport and Computing
Platform”, 5G-TRANSFORMER Deliverable D2.1, March 2018
[3] T. Deiss et al., “Definition of vertical service descriptors and SO NBI”,
5G-TRANSFORMER Deliverable D3.1, March 2018
[4] A. Garcia-Saavedra et al., “Definition of service orchestration and federa-
tion algorithms, service monitoring algorithms”, 5G-TRANSFORMER Delivera-
ble D4.1, March 2018
5G Research in Nextworks
Nicola Ciulli 23
Abstract The deployment and availability of 5G technologies are going to have

an impact on the way we develop and market our current product lines in the area
of Building Management Systems, Smart Buildings, and on the future ones in
Smart Factory.
1 5G impact on our products and business
During its lifetime, Nextworks has developed and commercialized a range of in-
novative products and solutions in the domain of facility control and automation,
from luxury yachting to residential / office building. The current focus of Next-
works’ product lines is in Building Management Systems, and solutions for Smart
Buildings and Smart Neighborhoods at large. These encompass divers systems for
comfort, entertainment, network, security, remote management, energy manage-
ment, works management and communications.
The advent and spreading of 5G technologies will have a strong impact on
our production activity in at least to major ways: in the way we conceive and
develop our technology solutions, and in the way we bring them to the market (i.e.
business models and go-to-market strategies).
Nicola Ciulli
Nextworks
446 Nicola Ciulli
Concerning the technology aspects, 5G and its architectural enablers (e.g.

SDN, NFV, MEC) are expected to have an impact on several subsystems in our
Smart Building solution.
Virtualization of Smart Building functions. Some years ago we have started
an R&D thread on the virtualization of some of the BMS functions, to run them
outside the technical facilities located in the building (e.g. in the cloud), with the
goal to maximize the utilization of IT resources and streamline building manage-
ment operations in those cases where several buildings are under the same own-
ership or control. However, some of these functions (e.g. those related to lighting
or other subsystems with hard real-time requirements) cannot deliver the right
performance behaviors when running in the cloud. A mixed cloud/local virtual-
ization approach provides the necessary architectural fix, and the adoption of Mo-
bile Edge Computing principles are helping us implement it.
Indoor communication infrastructure. Small cell 5G solution can provide the
next generation of building networking solutions, which are specifically important
for the refitting of existing buildings (e.g. historical buildings, where a wired net-
work cannot be deployed). Thanks to the availability of a mini C-RAN fully
fledged radio architecture with the related benefits (e.g. MEC), 5G for indoor
communication has the chance to surpass the solutions that are currently adopted,
which are based on proprietary wireless automation solutions and/or Wi-Fi.
A second line of impact from the adoption of 5G on our business is our de-
velopment of smart factory product lines. The technologies behind our current
BMS / Smart Building products can be naturally and easily evolved towards so-
lutions for some of the use cases in the factory of the future, in particular, those
related to overall facility automation (lighting, HVAC), shopfloor automation and
logistics in general. Our current plans already include this evolution, and the
knowledge that we have developed in 5G control plane enabling architectures and
technologies (SDN, NFV and MEC) can be instrumental in two ways: allow us to
develop/integrate the necessary technologies, and help build our image in a brand
new market domain.
Open issues and beyond 5G
Abstract
Fifth generation mobile networks (5G) will change our life and society, thanks to
the introduction of recently developed technologies which will enable the develop-
ment of new services and applications. While 5G is appearing with her novelties,
we need to start thinking about the next steps that wireless cellular systems have
to tackle, always keeping in mind that the final goal is the benefit of humanity. In
this chapter we discuss some issues that 5G will still leave open, and the possible
evolution towards the next generation (6G) of wireless communication systems.
1 Introduction
The last two decades have witnessed an extremely fast evolution of mobile cellular
network technologies, from 1G to 4G, with 5G networks expected to be operational by
2020. This chapter will briefly review the path from the first to the latest generation of
mobile cellular systems, will discuss some issues not fully addressed in 5G systems,
and finally will sketch a vision of what we may expect beyond 5G. For the path
beyond 5G, we envision that the mobile network will become more intelligent,
with learning mechanisms to modify itself based on users’ experience; situation-
awareness will lead decision making and networking; this will allow fast and flexible
spectrum reallocation, with consequent large bitrates available to the users; other
human senses will be communicated, and 3D / holographic type communication
will improve the quality of the tele-interaction; users will not necessarily need to
bring a smartphone but will benefit of wireless-devices-as-a-service, with distributed
devices available to anyone; the devices battery life will be substantially extended.
Among the technologies to reach these goals we count machine learning, dynamic
spectrum allocation, wireless energy transfer, free-space optical communication, use
of bands beyond 100 GHz, massive use of multiple antenna systems, new access
CNIT/University of Bologna, e-mail: {marco.chiani, e.paolini, franco.callegati}@unibo.it
447
448 Marco Chiani, Enrico Paolini, Franco Callegati
Fig. 1 The network will learn from experience.
schemes allowing an increase in node density, cybersecurity for quantum attacks,

high accuracy indoor localization, massive virtualization of network functionalities
and novel, software driven, network control architectures.
2 The evolution from 1G to 5G
Table 1 provides an overview of the main features of each cellular network generation,
from the first generation to the fifth one, including regulation, services, innovation
with respect to the previous generation, and some ancillary information.
Concerning regulation, we witnessed a profound paradigm shift across the several
generations, from 1G state-owned monopoly operators, very often obtaining the
use of spectrum free of charge, to open-market auctions starting with 3G, to 5G
spectrum sharing. Concerning open-market auctions, 3G systems spectrum was
licensed at extremely high prices, with up-front payments years before the first
universal mobile telecommunications system (UMTS) terminal appeared on the
Open issues and beyond 5G 449
market. This regulation model contributed to late (and in some cases limited) profits
seen by operators and, consequently, to a limited availability to excessive up-front
payments for 4G spectrum licenses. The possibility of extending spectrum licenses
up to 25 years, to apply spectrum sharing mechanisms, and to benefit from some
regulatory flexibility in new millimeter wave (mmWave) frequency bands, are new
regulatory aspects characterizing 5G.
In terms of services, the initial basic voice-only calls featured by 1G mobile sys-
tems evolved into a multitude of different services with the subsequent generations,
from simple text messaging and basic high-latency data exchange to high-quality
video streaming and chatting services, to radically new services supported by 5G
networks. Services evolution was enabled by several factors, including ever-rising
supported bit rates, advances in air interface design, signal processing at physi-
cal layer, and MAC layer procedures, technological advances in mobile terminals
manufacturing, evolution of mobile internet protocols, cloud computing, advanced
networking control paradigms.
Second generation global system for mobile communications (GSM) cellular
networks initially provided digital voice service at bit rate 9.6 kbps. General Packet
Radio Service (GPRS) and ultimately Enhanced Data Rates for GSM Evolution
(EDGE) data services were subsequently introduced, with bit rates of a few tens
of kbps and up to 200 kbps, respectively. Push email was also introduced for the
first time on Blackberry devices. These bit rates were largely increased in next
generations. Third generation UMTS offered up to 2 Mbps bit rate (often 364 kbps)
initially, and then several tens of Mbps in downlink with High Speed Packet Access
(HSPA). Fourth-generation LTE features up to 300 Mbps in downlink, with a target
of 1 Gbps, and up to 50 Mbps in uplink. Fifth generation cellular networks are
expected to increase the bit rate significantly, up to 20 Gbps. These bit rates, end-
to-end latencies down to 1 ms, ultra reliability (packet error rate 10−5 or less), and
massive multiple access, will foster services such as enhanced mobile broadband,
device-to-device (D2D) communication, ultra-reliable and low-latency Internet of
Things (IoT) and machine-type communication (MTC), e-health, augmented reality
and tactile Internet, industrial control for the Industry 4.0, automated driving and
flying. D2D communication, consisting of establishing a direct link between nearby
devices without relaying information through a base station (BS), is emerging as a
key technology to achieve efficient resource allocation, higher spectral efficiency,
reduced latency.
Each generation of wireless cellular networks came with its own technical inno-
vations, both on the network, air interface, and user terminal side.
Multiple access. Multiple access schemes were constantly enhanced, from sin-
gle carrier per channel frequency division multiple access (FDMA), to frequency-
and-time division multiple access (FDMA/TDMA), code division multiple ac-
cess (CDMA), up to orthogonal frequency division multiple access (OFDMA) and
non-orthogonal multiple access (NOMA), with a progressive explosion of the net-
work capacity in terms of the number of users served at the same time with an
adequate quality of service (QoS).
PHY layer enhancements. Fundamental innovations at the physical layer con-

tributed to the above-reviewed ever-increasing bit rates. Among them we can count
multiple-input multiple-output (MIMO), evolving into massive MIMO in 5G, as well
as advanced channel coding schemes such as 3G/4G turbo codes and 5G New Radio
traffic channel low-density parity-check (LDPC) codes and control channel polar
codes.
Frequency bands New frequency bands were exploited by each generation. First
generation advanced mobile phone system (AMPS) and total access communication
system (TACS) were operated in the 800 MHz and 900 MHz bands, respectively,
while second generation GSM was initially operated in the 900 MHz band, and
then also in the 1800 MHz and 1900 MHz (in North America) ones. Frequency
bands around 2 GHz were for the first time used by UMTS networks, while a
number of frequency bands are available worldwide for long term evolution (LTE),
based on regulatory aspects in different geographical areas (e.g., 450/800/900 MHz,
1800/2100 MHz, 2600 MHz in Europe). From a spectrum allocation viewpoint, the
main breakthrough introduced by 5G is the use of licensed, shared, and unlicensed
frequency bands in the mmWave band, above 24 GHz. The one around 60 GHz is
of particular interest for indoor very-high data rate applications, wireless backhaul,
and femtocell implementation.
Evolution of switching and networking The 1G network was fully circuit based,
resembling the traditional telephone network, a choice motivated by being voice
calls the target service of the networks. In 2G and 3G systems data oriented services
progressively became more important and a packet switched network dedicated
to the transport of data was set aside the existing circuit switched network. From
4G the evolved packet core (EPC) concept is adopted meaning that the transport
network is based on packet switching only and that such packet switching is IP
based. Therefore the transport technology of the mobile networks converges with
the transport technology of the Internet. The EPC definition brings along another
concept fundamental for the evolution to the 5G: the separation of the control and
data plane. The technologies to control the transport of the user data are logically
separated from the transport itself, which is essential to allow a separate evolution
path to guarantee the required network scalability.
Network softwarization The emergence of the cloud computing paradigm and
of the related virtualization technologies brought forward a key innovation in the
5G. Complex networking functionalities, traditionally requiring dedicated hardware
and management may be virtualized as pieces of software into the cloud. This is the
network function virtualization (NFV) paradigm, defined and standardized by ETSI
in the recent years. Meanwhile software defined networking (SDN) also emerged,
a novel approach that allows a very effective separation of the network control and
data plane, thus further extending the ideas behind the EPC. The OpenFlow protocol
is the key component of SDN, supporting the implementation of a communication
channel between the controller (the brain of the network) and the network nodes
(that carry data). NFV exploits virtualization, cloud computing and SDN to define
an architecture which supports the implementation of network services as subsets of
software functionalities, with an unprecedented degree of flexibility and adaptation
capabilities. Network programmability and split of control and data plane are key
enabling technologies towards a full implementation of the 5G. NFV and SDN fulfill
such expectation and are now getting into a state of advanced validation.
3 5G Issues
This section points out some issues and challenges that, at present, have not been
fully addressed in 5G mobile networks and that are attracting further research efforts.
Coverage issues. With the advent of 5G we will witness a site densification pro-
cess as a primary means to increase network capacity. Site densification necessarily
poses economical issues that may slow down considerably spatial and temporal 5G
deployment, unless a substantial BS cost reduction is achieved over the time. It has
been reported how SDN and network virtualization may contribute to cut costs,
but there remains uncertainty about to what extent this will speed up 5G rollout.
Recent studies have shown that, under a business-as-usual model, in UK 90% of
the population will be covered with 5G not before 2027 and the that 100% coverage
will be extremely hard to reach due to prohibitively increasing deployment costs in
less populated areas [7]. Similar expectations have been reported for other countries
during discussions at the 2018 IEEE 5G World Forum.
Emerging applications challenges. Super-hype of 5G contributed to create un-
precedented expectations about the levels of QoS these mobile networks will be
able to provide. It is becoming clear, however, how some emerging applications will
push the required QoS to extreme levels that appear very challenging for currently
envisioned 5G architectures. Among them, driverless cars and vehicles and the tac-
tile Internet. For example, the set of requirements imposed by the tactile Internet
(end-to-end ultra-low latency not exceeding 1 ms, outage probability 10−7 or less,
network intelligence to support predictive actuation) to deliver actuation and senses
such as hearing, touching, and seeing, is still considered a 5G challenge especially
over long ranges beyond 100 km [9].
D2D challenges and vulnerabilities. There are aspects of D2D communication
in 5G that have not yet been addressed in a totally satisfactory manner. One is coex-
istence of cellular users (CUs) and D2D pairs, particularly, mitigation of D2D links
interference on CUs, when downlink resources are shared with D2D devices, and
also on the BS when uplink resources are shared. D2D interference management is
still a subject of research efforts, since existing interference mitigation techniques
(interference cancellation based on coding and signal processing, interference avoid-
ance based on orthogonal resource allocation, and interference coordination based
on scheduling and power control) are expected to be insufficient in ultra-dense node
deployment scenarios. Other issues concern D2D security and privacy. Direct or
relay-assisted communication may be established on device controlled links with
no control of the core network, a trusted party providing identification, authentica-
tion, and encryption. This makes D2D links potentially more vulnerable to privacy
violation, besides suffering from all of the attacks affecting other networks.
Mobile edge computing issues. It is foreseen that moving computing, storage,

and networking resources to the edge of the radio access network (RAN) will be
a key ingredient to alleviate backhaul and core network and to allow executing
delay-sensitive and context-aware applications in the proximity of end users. This
paradigm, referred to as mobile edge computing (MEC), poses however concerns
[10]. Among them, the limited computing and storage resources per each MEC
platform, the necessity for MEC platforms of different provider to collaborate, chal-
lenges in user mobility support in small cells, problematic applicability of centralized
authentication protocols.
An open and smart RAN? We are currently witnessing efforts towards open,
interoperable interfaces, RAN virtualization, and RAN intelligence.There is a trend
to incentivize the use of commercial off-the-shelf (COTS) hardware, to explore open
source and open whitebox network by introducing virtualized network elements
with standardized open interfaces, to push embedded machine learning systems and
artificial intelligence back end modules for an enhanced network intelligence. These
are the main objectives of the ORAN Alliance. This trend is taking the first steps and
it is clear to what extent the envisioned features will be implemented in 5G networks.
Network orchestration and slicing. The NFV-SDN technologies, as briefly men-
tioned above, promise to support the implementation of a large variety of services
by means of softwarized functionalities hosted into cloud computing platform over
data centers implemented with standard hardware (COTS). In terms of Capex and
Opex this is a real revolution for network operators, that can move into the direc-
tion of more effective procurement and simplified operations. Moreover the same
resources (both and hardware and software) may be shared among different subsets
of customers, while keeping full isolation of the data paths and of the quality of
service thanks to the native capabilities of cloud computing platforms. This is the
"slicing" concepts, different subsets of customers may subscribe a service contract
with different operators (either real or virtual) and share in the end the same infras-
tructure, which paves the way to novel business models and opportunities for the
network providers. An effective management of NFV requires an orchestration plat-
form that will automate the deployment of the required functionalities and manage
operations. Some opensource implementations of such an orchestration platform are
currently under implementation, for instance Open source MANO (OSM)1, and the
open network automation platform (ONAP)2, and running proof of concepts are also
available such as the central office re-engineered as a data-center (CORD)3. These
technologies are not at the production level and some issues like multi-domain or-
chestration support are still open. Similarly, the slicing paradigm has still to be fully
understood and experimented to validate some critical aspects such as full quality of
service isolation and security guarantees.
1 https://osm.etsi.org
2 https://www.onap.org
3 https://opencord.org
4 A Vision of 6G
Some possible facts which will arise are summarized below.

More and more data. As of today (2018), about one million Terabytes (i.e.,
1 Exabyte = 1018 bytes) of data per day are exchanged over the mobile networks
all over the world (see Fig. 2). The amount of data exchanged by mobile users will
continue to increase, on one side due to the increasing number of non-human devices
connected (including vehicles, UAVs, and autonomous systems), and to the enhanced
quality 3D video / holographic type communication that will be used by humans.
Network Intelligence. To deal with increased traffic, the mobile network will
become more intelligent, with learning mechanisms to autonomously modify itself
based on users’ experience, and situation-aware decision making and networking.
Fast spectrum reallocation. Network intelligence will be used to allow fast and
flexible spectrum allocation/reallocation, with consequent large bitrates available to
the users.
Enhanced senses. Other human senses will be communicated to improve the
quality of the tele-interaction, including 3D / holographic type communication,
taste, smell, touch.
Wireless-devices-as-a-service. Users will not necessarily need to bring a smart-
phone but will benefit of wireless-devices-as-a-service, with distributed devices
available to anyone. All information being in the cloud, users will just need to be
authenticated and then access the network by using any available device.
Battery duration & energy. The need to put devices on recharge will be dramat-
ically reduced, so that the battery life will be substantially extended.
Quantum computers & quantum networks. We will see the appearance of
quantum computers, capable to solve problems that are not solvable with non-
quantum computers. Also, quantum communication and networks will be available,
e.g., for cryptographic key exchange, also from satellites.
Privacy, security, data manipulation. The need for privacy and security for a
proper management of personal data will be of paramount importance. The appear-
ance of quantum computers will force to re-think the cryptography and security
mechanisms.
Security and safety. The IoT and the Industry 4.0 will bring the network very
close to the real infrastructures. Therefore a security breach on the networking side
may quickly become a very important safety issue in the real life. This already
happened with the failed attack to a petrochemical plant in Saudi Arabia in August
2017, that is widely believed was aimed to make the plant explode by attacking the
IT control infrastructure.
Virtual Operators explosion. The NFV-SDN technologies and the related slic-
ing capabilites will boost the emergence of virtual operators that will compete by
providing added values services to the users by deploying innovative network func-
tionalities in the cloud. This may lead to a significant innovation also in the business
models and commercial strategies in the field.
5 Technologies
Among the technologies in the evolution path, we cite the following.

Machine Learning & AI. The complexity of the network and the number of
connected devices will lead to a network which will learn from the experience to
modify itself and accomodate new services. A suitable policy should be investigated
to avoid the “Skynet” of the movie “The Terminator”.
Dynamic spectrum allocation. The precious radio spectrum in the lower bands
will be used more effectively by allocating the frequencies every second or so, based
on the context.
Wireless energy transfer. This could be in some situations a viable way to extend
the battery life, avoiding frequent recharges.
Free-space optical communication. Optical free-space communication will al-
low high data rate, for both outdoor and indoor scenarios, so releasing the lower
band radio spectrum for long range uses.
Sub-Terahertz and Terahertz communications. The availability of large bands
beyond 100 GHz (D-band, 110 GHz to 170 GHz,...) will lead to transmission systems
at high data rate over short distances, so releasing the lower band radio spectrum for
long range uses (see Fig. 3).
Massive use of multiple antenna systems. The increase in the frequency will
require multiple antenna systems able to exploit the multi-rays propagation, for larger
throughput but also for precise localization and for energy transfer.
High accuracy indoor localization. Context awareness needs a precise user’s
localization. For this item see the chapter about localization in this book.
Access schemes for massive wireless networks. New access schemes will
be needed to handle a massive number of non-orthogonal users (more than 10
devices/m2 ) in an efficient and scalable way.
Cybersecurity. The possibility to use non-personal devices for personal commu-
nication will impose new challenges on biometric authentication and privacy. New
schemes for wireless security, capable to deal with attacks of quantum computers,
will be needed.
References
1. IEEE Future Networks Initiative. URL: https://futurenetworks.ieee.org/.

2. Ericsson Mobility Report, Q2, 2018. URL: https://www.ericsson.com/en/mobility-report.
3. IEEE 5G World Forum, July 2018. URL: http://ieee-wf-5g.org/past-events/2018-2/.
4. ITU Focus Group on Technologies for Network 2030. URL: https://www.itu.int/en/ITU-
T/focusgroups/net2030.
5. K. David and H. Berndt. 6G Vision and Requirements: Is There Any Need for Beyond 5G?
IEEE Vehicular Technology Magazine, 13(3):72–80, 2018.
6. S. Kim, W. T. Khan, A. Zajić, and J. Papapolymerou. D-Band Channel Measurements and
Characterization for Indoor Applications. IEEE Transactions on Antennas and Propagation,
63(7):3198–3207, July 2015.
7. E. J. Oughton and Z. Frias. The cost, coverage and rollout implications of 5G infrastructure in
Britain. Telecommunications Policy, 42(8):636–652, 2018.
8. P. H Pathak, X. Feng, P. Hu, and P. Mohapatra. Visible light communication, networking,
and sensing: A survey, potential and challenges. IEEE communications surveys & tutorials,
17(4):2047–2077, 2015.
9. M. Simsek, A. Aijaz, M. Dohler, J. Sachs, and G. Fettweis. 5G-enabled tactile Internet. IEEE
J. Sel. Areas Commun., 34(3):460–473, March 2016.
10. T. X. Tran, A. Hajisami, P. Pandey, and D. Pompili. Collaborative mobile edge computing in
5G networks: New paradigms, scenarios, and challenges. IEEE Commun. Mag., 55(4):54–61,
April 2017.
11. Y. Xing and T. S Rappaport. Propagation Measurement System and Approach at 140 GHz-
Moving to 6G and Above 100 GHz. arXiv preprint arXiv:1808.07594, to appear in IEEE
Globecom 2018, 2018.
456
Generation Regulation Services Innovations Notes

1G monopoly voice lightweight mobile terminals humans, 10% popul.
2 G / GSM competing operators cho- voice, SMS, data 9.6 − 200 FDMA/TDMA, smaller mobile humans, > 90% popul., M2M
sen by regulators kbps (GPRS-EDGE), push email terminals, first cameras, GPS,
(Blackberry) content (iPhone), access to
WWW (WAP, not great succ.)
3 G / UMTS granting licenses, open- high data rates 364 kbps - 10 CDMA, camera, GPS superhype, limited success due to
market auction Mbps, email, file/image competition with cheaper WiFi,
mobile devices too heavy, low bat-
tery duration compared to GSM
4 G / LTE less up-front money from downlink ∼ 300 Mbps, uplink up OFDM, MIMO, D2D competitive with WiFi, elimina-
operators, elimination of to ∼ 50 Mbps, mobile video, mul- tion of roaming charges in Europe
roaming charges in Eu- timedia messaging (WhatsApp,
rope Instagram, Telegram, Twitter,...)
5G spectrum licenses up to bit rates ∼ 20 Gbps, lower laten- frequencies up to 60 GHz for superhype
25 years, spectrum shar- cies ∼ 1 ms, D2D, massive M2M extremely high-speed transmis-
ing, reuse, new frequency for IoT sion, massive MIMO, LDPC, Po-
bands mmWave lar Codes, network function vir-
tualization, software-defined net-
working, network slicing
Table 1 Evolution from 1G to 5G

Traffic growth is driven by both the rising percentage change for mobile data.
number of smartphone subscriptions This curve illustrates a typical moderating
and expanding average data volume per growth rate as the market evolves over time.
subscription, fueled primarily by more In Q2 2018, mobile data traffic grew
viewing of video content at increasingly by around 52 percent year-on-year and
higher resolutions. 9 percent quarter-on-quarter. There are
The graph below shows total global large differences in traffic levels between
monthly data and
Open voiceand
issues trafficbeyond
from Q2 2013
5G markets, regions and operators. 457
to Q2 2018, along with the year-on-year
Voice Data Year-on-year growth
18 180
16 160
14 140
Total (uplink + downlink) traffic (exabytes per month)
12 120
Year-on-year growth (percent)

10 100
8 80
6 60
4 40
2 20
0 0
Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2
2013 2014 2015 2016 2017 2018
Source: Ericsson traffic measurements (Q2 2018)
Fig. 2 Mobile data

Ericsson traffic (doesThe
notcontent
include DVB-H,
of this documentWi-Fi)
is subject[2].
to EAB-18:008331 Uen, Revision A
SE-164 80 Stockholm, Sweden revision without notice due to continued © Ericsson 2018
Telephone +46 10 719 0000 progress in methodology, design and
www.ericsson.com manufacturing. Ericsson shall have no
liability for any error or damage of any kind
10 2 resulting from the use of this document
10 1
Specific Attenuation (dB/km)
10 0
10 -1
-2
10
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300
Frequency (GHz)
Fig. 3 Specific attenuation of atmospheric gases at sea-level.

5G Italy White Ebook

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

5G Italy White Ebook

Uploaded by

Copyright:

Available Formats

i

5G Italy White Book:

All works of authorship contained in this publications are either original to

5G is the new generation of the global telecommunication network. It is

- network softwarization, i.e., the virtualization of network services, obtained

and key characteristics of this game-changing technology, proceeds to present

Roma, The eBook Editors

*CNIT (National, Inter-University Consortium for Telecommunications,

Our society is undergoing a rapid and profound transformation; a digital

The 5G for Europe Action Plan

The 5G Observatory – monitoring the position of

Europe’s competitive advantage

Europe’s approach to 5G deployment is also coordinated in other ways. While

The 5G Corridors for Connected and Automated

As a fully connected digital society fueled by 5G comes ever closer to becom-

Bruxelles, November 2018

2 Leading-edge 5G Research and Innovation: An undivided

Part II What is 5G as of today

9 Post-OFDM modulations for 5G and beyond . . . . . . . . . . . . . . . . . . . . 109

20 The role of propagation software tools for planning 5G wireless

22 Impulse Radio NOMA Communications in 5G and Beyond . . . . . . . . 277

Part III 5G research in Italy

24 Contributions to EU Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299

25 Open issues and beyond 5G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 447

Francisco Fontes, Ioannis Neokosmidis, Riccardo Trivisonno, Franco Davoli, Le

provide significantly better performance and new capabilities, mandatory for

1. 5G, the enabler of new services

4G was designed to improve capacity, user data-rates, spectrum usage and

“Enabling a seamlessly connected society in the 2020 timeframe and

In fact, 5G is designed to create the conditions to launch new applications

Figure 1: 5G new service capabilities

As a conclusion, 5G will efficiently support the three mentioned different

network needs and contribute to the digitalization of vertical markets such as

2. 5G as a promoter of a new ecosystem

5G can be thought of as being more like a Business-to-Business (B2B) rather

The ongoing discussions in standards on the next features to be added to the

with a rollout of software solutions from their in-house development labs, or

5G is mandated to significantly increase the available communication system

services. Furthermore, allowing for even more advanced sharing infrastructure

Figure 2: IMT-2020, enhancement of key capabilities [1]

It is important to highlight that not all of the enhanced capabilities described

4. Innovation, at the heart of 5G

To achieve the expected enhancements in capacity, coverage, reliability, latency,

A new 5G Core to interconnect everything

The architectural changes to the telecommunication systems, needed to fulfil the

These characteristics allow 5G deployments to adapt to a wide range of

A new radio interface to connect everything

A wireless network made of optics

With the advent of the 5G era, requirements like CapEx reduction,

Centralized Units (DU/CU), in a balanced splitting of functions, emerges as

Software based architecture

5G will be driven by software. Network functions are expected in a number

Nodes and in the end user premises. The 5G architectural framework is

Artificial Intelligence powering 5G

5. 5G Standardization and launch of first commercial

Several standards bodies are involved in the definition of the forthcoming 5G

Bernard Barani and Peter Stuckmann2

2 5G Vision driving R&D and technological

IMT-ADVANCED IMT-2020 (5G)

DL: 1 Gbps DL: 20 Gbps

PEAK SPECTRAL DL: 15 bps/Hz DL: 30 bps/Hz

MOBILITY 350 km/h 500 km/h

BANDWIDTH UP TO 20 MHz/channel Up to 1 GHz/channel