Professional Documents
Culture Documents
5G Italy White Ebook
5G Italy White Ebook
Edited by:
Marco Ajmone Marsan
Nicola Blefari Melazzi
Stefano Buzzi
ii
ISBN: 9788832170016
Preface
iii
iv Preface
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
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.
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.
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.
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
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
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
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
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.
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.
3. 5G enhanced performance
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
12 F. Fontes, I. Neokosmidis, R. Trivisonno, F. Davoli, Le Binh,
S. Mikroulis, I. Tomkos, V. Frascolla
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.
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.
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
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.
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.
USER
EXPERIENCED 10 Mbps 100 Mbps
RATE
20 Bernard Barani and Peter Stuckmann
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
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"
Leading-edge 5G Research and 21
Innovation: An undivided
commitment of Europe
Metz-Merzig-Luxembourg: FR-DE-LU
Rotterdam-Antwerpen-Eindhoven: NL-BE
EE-LV-LT Via Baltica (E67) Tallinn (EE) – Riga (LV) – Kaunas (LT) –
Lithuanian/Polish border
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
Leading-edge 5G Research and 23
Innovation: An undivided
commitment of Europe
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
Leading-edge 5G Research and 27
Innovation: An undivided
commitment of Europe
Phase one projects contributed to 15 core areas of innovation xxv, the “Golden
Nuggets”, hereafter briefly reported.
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
Leading-edge 5G Research and 31
Innovation: An undivided
commitment of Europe
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
32 Bernard Barani and Peter Stuckmann
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.
Leading-edge 5G Research and 33
Innovation: An undivided
commitment of Europe
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
Leading-edge 5G Research and 35
Innovation: An undivided
commitment of Europe
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
36 Bernard Barani and Peter Stuckmann
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 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
Leading-edge 5G Research and 39
Innovation: An undivided
commitment of Europe
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
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
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.
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.
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
Leading-edge 5G Research and 45
Innovation: An undivided
commitment of Europe
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
46 Bernard Barani and Peter Stuckmann
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).
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
Leading-edge 5G Research and 49
Innovation: An undivided
commitment of Europe
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:
• 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
Leading-edge 5G Research and 51
Innovation: An undivided
commitment of Europe
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
52 Bernard Barani and Peter Stuckmann
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
https://5g-ppp.eu/5g-ppp-phase-2-projects/
xxix
http://5gobservatory.eu/5g-spectrum/national-5g-spectrum-assignment/
xxx
https://5g-ppp.eu/5g-ppp-phase-3-projects/
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
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.
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.
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 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])
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
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
Functional Architecture 63
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
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.
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.
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
1 Introduction
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.
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
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
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].
72 Michele Polese, Marco Giordani, Michele Zorzi
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].
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.
3GPP NR: the standard for 5G cellular networks 73
frequency
Synchronization Signal (SS) block
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
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
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].
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.
76 Michele Polese, Marco Giordani, Michele Zorzi
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.
Fig. 4 Representation of different deployment and operation modes for the 5G NR RAN and the
5GC.
3GPP NR: the standard for 5G cellular networks 77
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.
78 Michele Polese, Marco Giordani, Michele Zorzi
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
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
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
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.
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.
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
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
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.
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.
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].
Use of millimeter wave carrier frequencies in 5G 85
y1 s1
Analog Beamforming
Digital Beamforming
DAC
d1
dM
y M̃ s NR
DAC
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.
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-
Use of millimeter wave carrier frequencies in 5G 87
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.
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
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;
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].
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.
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.
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
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
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.
70
M-MMSE
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
• 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
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
• 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.
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.
Overview on Optical Fronthauling Technologies for Fixed-Mobile Convergence 103
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.
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)
Overview on Optical Fronthauling Technologies for Fixed-Mobile Convergence 105
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].
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
Overview on Optical Fronthauling Technologies for Fixed-Mobile Convergence 107
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
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.
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
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
112 Paolo Banelli, Giulio Colavolpe, Luca Rugini, Alessandro Ugolini
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.
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,
Post-OFDM modulations for 5G and beyond 113
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-
Post-OFDM modulations for 5G and beyond 115
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.
116 Paolo Banelli, Giulio Colavolpe, Luca Rugini, Alessandro Ugolini
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
Post-OFDM modulations for 5G and beyond 117
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.
118 Paolo Banelli, Giulio Colavolpe, Luca Rugini, Alessandro Ugolini
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.
120 Paolo Banelli, Giulio Colavolpe, Luca Rugini, Alessandro Ugolini
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
121
122 Giacomo Verticale, Antonio Capone
Service
consumer R R
R
SDN Controller
R Resource
R
Service data group
R
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
3 Cloud-RAN
4 Network Slicing
Services
Service Layer
Physical Virtualized
Infrastructure Infrastructure
Resource Resource
Slice with unaware unaware
Slice with shared slice slice
Standalone own resources
slice spectrum
Common PDCP / RLC
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.
Softwarization and Virtualization 127
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
128 Giacomo Verticale, Antonio Capone
References
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
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.
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
132 Luca Chiaraviglio, Michela Meo
STORAGE
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
Efficiency/Off Grid Operation 133
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
3 Cost Analysis
CAPEX [Euro]
6 Site Acquisition
0
UAV − Italy LC − Italy UAV − Cook Islands LC − Cook Islands UAV − Zimbabwe LC − Zimbabwe
Strategy − Scenario
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
136 Luca Chiaraviglio, Michela Meo
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
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
Efficiency/Off Grid Operation 137
4
10 Site 9
6
Site 10
Site 15
5
1
100 200 300 400 500 600 700
TS Index
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.
138 Luca Chiaraviglio, Michela Meo
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
Efficiency/Off Grid Operation 139
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].
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.
140 Luca Chiaraviglio, Michela Meo
Fig. 11 Daily profiles of renewable energy production per 1 kW p in different sample months in
Torino (results obtained with PVWatts) [9].
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
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].
Average battery charge [kWh]
30 30
All Days
Type 1
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.
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
Efficiency/Off Grid Operation 143
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
Giuseppe Bianchi
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.
Fig. 1 4G security architecture: roman numbers on top of each arrow specify which security domain
is involved for the considered interface.
2 5G security: vision
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.
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.
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
152 Giuseppe Bianchi
Cloud Server
Physical Core
L1 Cache
L2 Cache
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.
154 Giuseppe Bianchi
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
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.
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
functions and their relationship with respect to the ETSI NFV architectural
framework.
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 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
Reliable Slicing in 5G Networks 163
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
Reliable Slicing in 5G Networks 165
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).
166 Valcarenghi, Giorgetti, Martini, Kondepu, Gharbaoui, Castoldi
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
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.
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
• 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
• 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.
• 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.
170 Stefania Bartoletti, Andrea Conti, Davide Dardari, and Andrea Giorgetti
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.
5G Localization and Context-Awareness 171
Fig. 1 Potential requirements per use case highlighted by the 3GPP Rel. 16
172 Stefania Bartoletti, Andrea Conti, Davide Dardari, and Andrea Giorgetti
Accuracy
1Km
Long-range IoT CID
3G TDOA
100m E-CID
RFPM
10m 4G TDOA
A-GNSS
WLAN/BT
UWB
10cm
Coverage
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.
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
5G Localization and Context-Awareness 175
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
UMTS
BS 1
OTDOA BS1-BS3
hyperbola
PRS
BS 3
PRS
BS 2
PRS
RSTDs
Core network
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.
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
5G Localization and Context-Awareness 177
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.
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.
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.
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].
In-sensor processing
− Demodulation
− ToA estimation
− Clutter removal
− Ghost mitigation
− CFAR detection
− 1D clustering
− Measurement select.
y Objects trajectory
Network processing
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
5G Localization and Context-Awareness 181
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].
182 Stefania Bartoletti, Andrea Conti, Davide Dardari, and Andrea Giorgetti
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.
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.
5G Localization and Context-Awareness 185
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.
186 Stefania Bartoletti, Andrea Conti, Davide Dardari, and Andrea Giorgetti
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
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.
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
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.
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
192 Roberto Verdone, Silvia Mignardi
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.
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.
194 Roberto Verdone, Silvia Mignardi
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.
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.
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.
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.
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
1 Introduction
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
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.
202 Fabrizio Granelli
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.
Fig. 4 Potential energy gains deriving from the deployment of the Cloud RAN paradigm: in
presence of low traffic, fewer BBUs are required.
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-
The Role of Cloud and MEC in 5G 205
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
206 Fabrizio Granelli
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.
The Role of Cloud and MEC in 5G 207
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
1 Introduction
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
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
212 Antonella Molinaro and Claudia Campolo
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.
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
5G for V2X Communications 213
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.
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
214 Antonella Molinaro and Claudia Campolo
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-
5G for V2X Communications 215
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].
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.
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
5G for V2X Communications 217
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
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.
218 Antonella Molinaro and Claudia Campolo
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
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
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
2 Related Works
Figure 1 depicts the proposed DAVOSS architecture. The system can be divided into
four main layers:
222 Riccardo Bassoli, Fabrizio Granelli
• 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.
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;
Rapid Deployment of 5G Services Using Drones 223
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).
226 Riccardo Bassoli, Fabrizio Granelli
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
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.
228 Riccardo Bassoli, Fabrizio Granelli
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.
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).
230 Riccardo Bassoli, Fabrizio Granelli
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
231
232 Angela Sara Cacciapuoti, Luca Chiaraviglio, Gerardo Di Martino and Marco Fiore
1 Introduction
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.
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
234 Angela Sara Cacciapuoti, Luca Chiaraviglio, Gerardo Di Martino and Marco Fiore
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.
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.
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 [?].
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
236 Angela Sara Cacciapuoti, Luca Chiaraviglio, Gerardo Di Martino and Marco Fiore
Legacy
pre-5G
Site
New
5G-Site
Service Area
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., [?, ?])
5G Planning under EMF Constraints 237
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.
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
238 Angela Sara Cacciapuoti, Luca Chiaraviglio, Gerardo Di Martino and Marco Fiore
(a) 100% maximum input power (b) 75% maximum input power
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
5G Planning under EMF Constraints 239
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
-130 -130
-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
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.
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.
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.
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
5G Planning under EMF Constraints 243
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
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
Università degli Studi di Napoli Federico II
Giuseppe Ruello
Università degli Studi di Napoli Federico II
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
248 A. Iodice, D. Riccio, G. Ruello
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.
Role of propagation tools 249
(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).
250 A. Iodice, D. Riccio, G. Ruello
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.
Role of propagation tools 251
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.
252 A. Iodice, D. Riccio, G. Ruello
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.
Role of propagation tools 253
Antonella Bogoni
Scuola Superiore Sant’Anna - Pisa
Luca Potì
Photonic Networks & Tecnhologies Nat’l Lab – CNIT, Pisa
Giancarlo Prati
Photonic Networks & Tecnhologies Nat’l Lab – CNIT, Pisa
IRCPhoNeT, Pisa
Marco Romagnoli
Photonic Networks & Tecnhologies Nat’l Lab – CNIT, Pisa
256 A. Bogoni, L.Potì, G. Prati, M. Romagnoli
Fig 2.
Xhaul concept
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].
Photonics for 5G 259
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
260 A. Bogoni, L.Potì, G. Prati, M. Romagnoli
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
Photonics for 5G 261
νc
-p
F ctrl
DF DF
DF DF
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
262 A. Bogoni, L.Potì, G. Prati, M. Romagnoli
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.
Photonics for 5G 263
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
2.1 Silicon
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.
Photonics for 5G 265
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.
266 A. Bogoni, L.Potì, G. Prati, M. Romagnoli
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 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.
268 A. Bogoni, L.Potì, G. Prati, M. Romagnoli
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
Photonics for 5G 269
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
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.
270 A. Bogoni, L.Potì, G. Prati, M. Romagnoli
Fig 10.
Picture of the photonics-based PS, with the PCB (printed circuit board) hosting the PIC.
Inset: Microscope picture of the PIC
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.
Photonics for 5G 271
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.
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.
Fig. 12
a) Co-packaged multi-chip: Silicon PIC, silicon EIC and laser [23]. b) Wafer bonded laser [24].
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.
Photonics for 5G 275
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
1 Introduction
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
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]
280 Mai T. P. Le, Giuseppe Caso, Luca De Nardis, Maria-Gabriella Di Benedetto
Table 1: NOMA schemes proposed for the Rel-14 3GPP NR Study Item [3]
Code-domain
NOMA
Single-carrier Multi-carrier
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.
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.
Impulse Radio NOMA Communications in 5G and Beyond1 283
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].
284 Mai T. P. Le, Giuseppe Caso, Luca De Nardis, Maria-Gabriella Di Benedetto
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.
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.
Impulse Radio NOMA Communications in 5G and Beyond1 285
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].
286 Mai T. P. Le, Giuseppe Caso, Luca De Nardis, Maria-Gabriella Di Benedetto
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
Impulse Radio NOMA Communications in 5G and Beyond1 287
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
288 Mai T. P. Le, Giuseppe Caso, Luca De Nardis, Maria-Gabriella Di Benedetto
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
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.
290 Mai T. P. Le, Giuseppe Caso, Luca De Nardis, Maria-Gabriella Di Benedetto
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
1 Introduction
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.
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
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.
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
• 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 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
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.
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.
Ericsson has identified six industrial social and industrial applications [4] in which
5G will bring substantial revolutions. The following sections illustrate such
5G and Relevant Enabling 305
Technologies
applications detailing the beneficial impact of 5G and the relevant main technol-
ogy enablers.
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
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 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
• Roundtrip latency in 1 ms range
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
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.
5G and Relevant Enabling 309
Technologies
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
• Roundtrip latency in 1 ms range
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)
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.
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.
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.
5G and Relevant Enabling 315
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
5G and Relevant Enabling 317
Technologies
. 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.
5G-MoNArch
Enabling innovations Common
architecture
Architecture Secure and resilient Deployment
Sea port
instantiation network slice
Cloud-enabled building testbed
In the context of the project concept outlined above, the individual objec-
tives of 5G-MoNArch are detailed next.
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.
322 Mauro Renato Boldi and the Project Partners
• 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.
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
324 Mauro Renato Boldi and the Project Partners
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.
Promising technology solutions and use cases for beyond 2020 communication
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.
328 Mauro Renato Boldi and the Project Partners
and
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
Technology Components
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
330 Mauro Renato Boldi and the Project Partners
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.
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.
334 Mauro Renato Boldi and the Project Partners
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 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.
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 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.
Two Italian partners contributed to the Project: Telecom Italia (TI) and Azcom
Technology (Azcom).
342 Mauro Renato Boldi and the Project Partners
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
Figure 1-1: 5G MiEdge concept, use cases (left side) and key technologies (right side).
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
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,
358 Mauro Renato Boldi and the Project Partners
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
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
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
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
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,
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.
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.
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
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)
Xhaul
End-Point Plane
Xhaul
General Processing Plane
Xhaul
Interconnection Plane
1 Project Objectives
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
376 D. Siracusa, E. Salvadori, G. Landi, N. Ciulli, C. Casetti, A.
Di Giglio, P. Iovanna
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
1 Project summary
5G-PPP Phase 1
H2020 Call H2020-ICT-2014-2
Project duration: 1 July 2015 – 30 December 2017
Project budget: € 6.024.299,00
Requested EC budget: € 6.024.299,00
Project coordinator: VTT Technical Research Centre of Finland Ltd
2 Objectives
• 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.
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
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
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 duration: 1 July 2015 – 30 December 2017
Project budget: € 8.206.935,00
Requested EC budget: € 7.488.434,00
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
In this Section, we provide a short overview of the major contributions from the
Italian partners to the project.
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
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
3. 5G-EmPOWER - http://5g-empower.io/
5G-ESSENCE project
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.
Daniele Munaretto
ATHONET
Antonino Albanese
Italtel
1 Project summary
5G-PPP Phase 2
H2020 Call H2020-ICT-2016-2
Project duration: 1 June 2017 – 30 November 2019
Project budget: € 7.978.272,50
Requested EC budget: € 7.978.272,50
Project coordinator: OTE AE (Greece)
2 Objectives
• 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
Italian partners to the project.
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
4. 5G-EmPOWER - http://5g-empower.io/
5G-CARMEN project
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
5G-PPP Phase 3
H2020 Call H2020-ICT-18-2018
Project duration: 1 November 2018 – 31 October 2021
Project budget: € 18.566.123,75
Requested EC budget: € 14.960.887,26
Project coordinator: Fondazione Bruno Kessler (FBK)
2 Objectives
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.
In this Section we provide a short overview of the major contributions from the
Italian partners to the project.
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
396 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
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
398 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
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
5G-CARMEN Project 399
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
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
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. 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.
408 Gino Carrozzo, Paolo Cruschelli, Maria Rita Spada, Mauro
Di Bugno, Luca Vignaroli, Antonino Albanese, Andrea Mi-
chelozzi
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)
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
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
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.
5G-MEDIA: Programmable 415
edge-to-cloud virtualization
fabric for the 5G Media indus-
try
_______________________________________________________
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.
5G-MEDIA: Programmable 417
edge-to-cloud virtualization
fabric for the 5G Media indus-
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.
_______________________________________________________
References
[1] H2020 5G-MEDIA Project web site, [Online]. Available: http://www.5gmedia.eu/, Accessed on: Oct. 2018
418 Francesca Moscatelli, Giacomo Bernini, Pasquale Andriani
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
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.
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
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)
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.
also ensuring that relevant standardization bodies from the two involved indus-
tries are impacted.
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
• 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.
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
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
of the resources for each slice. Microservices and Containers technology are used
for all produced SW modules.
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
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.
442 L. Valcarenghi et al.
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.
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.
Nicola Ciulli 23
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
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
447
448 Marco Chiani, Enrico Paolini, Franco Callegati
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).
450 Marco Chiani, Enrico Paolini, Franco Callegati
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.
452 Marco Chiani, Enrico Paolini, Franco Callegati
1 https://osm.etsi.org
2 https://www.onap.org
3 https://opencord.org
Open issues and beyond 5G 453
4 A Vision of 6G
5 Technologies
References
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
18 180
16 160
14 140
Total (uplink + downlink) traffic (exabytes per month)
12 120
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
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)