Professional Documents
Culture Documents
I. I NTRODUCTION
Mobile data traffic is expected to grow exponentially in
the coming years [1][2][3]. To meet this coming demand,
broadly agreed upon design targets for 5th generation (5G)
mobile networks are ambitious, with proposed peak data rates,
network capacity, energy efficiency, and latency design goals
being orders of magnitude greater than current-generation
capabilities [4][5].
Fig. 1: Some pieces of the Internet. It is a network of networks.
One important research focus has been efficient manage-
Internet service providers (ISPs) are organizations that supply
ment of radio spectrum to maximize end system radio access
fundamental infrastructure needed to communicate over the
and frequency band utilization. While mobile broadband com-
Internet. Figure modified from [8].
munications spectrum has been allocated on an international
level and an overall 5G framework is in place [6][7], technolo-
gies underpinning future mobile radio access networks (RANs)
remain to be standardized. This reality continues to promote computer) networks, as evidenced by the hegemonic status of
research into finding a product-level agreement on a 5G radio the Internet. In a data network, communication is achieved
access infrastructure. by using computing devices to send digitized information
The purpose of all telecommunications networks is to estab- across physical transmission media using formal rules called
lish information exchange between two or more end systems, protocols. A ”nuts and bolts” view of the Internet is shown in
which are devices that sit on the edge of a network. These Fig. 1. Modern mobile (or cellular) networks are data networks
end systems allow end users to access a network. The most that connect to the Internet.
important telecommunications networks today are data (or If an analogy were to be drawn between a rudimentary
data network and the old game of tin can telephones, the end
This paper was submitted for review on Apr. 25, 2017. It is an independent systems would be the tin cans (using the computing power of
work with no sponsors or financial contributions. human users in this case), the transmission medium would be
M. Ljubicic is with the Department of Electrical and Computer En-
gineering, University of Utah, Salt Lake City, UT 84112 USA (e-mail: the wire connecting them, and the protocol would be the set
marko.ljubicic@utah.edu). of rules governing the communication between the people at
2
A. Distributed Spatial Databases to cope by deploying cell sites more densely. However, this
Wireless end systems listen for beacons in each frequency imposes high costs not only in terms of capital and operational
band and then associate with nearby base stations according to expenditures, but also in terms of increasingly chaotic station
various suitability criteria. Increased traffic demand requires placement and inter-cell interference. SoftRAN, a software-
increased numbers of access points and cellular frequency based solution to this problem, proposes that all base stations
bands, which makes this access model inefficient in the long- deployed in a geographical area should be abstracted as a
term. However, the proliferation of access points and cellular virtual big-base station which is made up of radio elements
bands provides the ability to locate an end system’s precise (the individual physical base stations) [18]. A logically cen-
geographic location [1]. Knowing the geographic location of tralized control plane manages all handovers (transferring
an end system with high precision is useful for scanning communication with a wireless device from one cell site to
nearby BSs to find suitable access points. BS and available another) and interference. SoftRAN abstracts the available
cellular band information can be aggregated and stored in radio resources as a 3D resource grid of space, time, and
central spatial databases that are maintained by mobile net- frequency slots and programs them through a centralized radio
work operators (MNOs). An arrangement like this can be access control plane [16][18].
considered an ”information service” (IS) architecture, where RadioVisor has been proposed as an extension to SoftRAN.
essential network information is provided to end systems in It functions as a slicing plane to enable sharing of network re-
what is essentially a client-server model [1]. sources [16]. Instead of leaving the management of the abstract
As can be seen in Fig. 6, the IS architecture works by aggre- radio plane to a single centralized controller, which limits
gating network information across multiple base stations into a sharing of network resources, RadioVisor divides 3D resource
distributed network of servers that functions as a single logical grid among virtual operators by introducing a slice manager
spatial database. End system queries about network state are and multiple independent controllers (one for each operator).
relayed to this database, which consists of root-level MNO This allows for much more dynamic resource allocation [16].
servers at the highest level, regional servers in the middle,
and possibly more localized servers at the bottom. Simulations C. Programmable Radio Access Networks (PRANs)
using MySQL and PostgreSQL show that a single central
Current Long Term Evolution (LTE) RANs consist of a
database can handle millions of queries per hour and provide
collection of largely independent base stations [17]. LTE is
nationwide information service [1]. This level of scalability
a widely deployed 4G wireless broadband technology [19].
makes it well-suited as a 5G radio access architecture.
It is an RAN technology, and similar to the protocol stack of
wired computer networks (and wireless LANs), it is composed
B. RadioVisor: Flexible Sharing of Radio Access Networks of layers with separate (logical) responsibilities. The top
The explosive growth in both the volume and diversity two layers of the LTE RAN, which are essentially TCP/IP,
of mobile network traffic has pressured wireless networks are programmable using modern software and cloud-based
6
transmitted to the network on the macrocell’s carrier fre- cellular architecture: an LSA/ASA controller and an LSA/ASA
quency. Fig. 10 shows a broad view of the SDRAN. repository [22].
The LSA/ASA repository is a database that stores the
III. S PECTRAL E FFICIENCY necessary shared spectrum information [22]. It also acts as an
interface between an incumbent spectrum user, who is using
Radio access concerns itself with efficient end system
the spectrum based on license-specific policies, a regulator,
access to cellular bands over a wireless network, but an-
who defines the rules, conditions, and licensing-terms of the
other important area of interest for 5G RAN performance
sharing framework, and the controller. Based on the informa-
is spectral efficiency: the amount of data bandwidth that a
tion it collects, the repository defines protected areas according
specific technology can extract from a certain amount of radio
to the underlying regulatory requirements. These protected
spectrum [21]. While traffic flow across the mobile spectrum
areas are exclusion zones where the LSA/ASA users are not
is vast, there is, at any moment in time, a large amount of
allowed to use the given LSA/ASA band [22].
idle bandwidth; there is also allocated spectrum that remains
The LSA/ASA controller provides an MNO a means to
available but unused for various reasons. To improve overall
obtain information about available spectrum bands and to
spectrum usage, spectrum sharing has been identified as a
access the spectrum [22]. The controller requests available
building block for 5G networks.
frequencies from the repository for a specific location. For
Spectrum sharing refers to a situation where two or more
efficient management of the MNOs network and spectrum
systems are operating in the same frequency band [22]. The
resources according to demand, the controller receives all
mobile spectrum can be divided into a licensed and an un-
the needed network related information from its network
licensed category, with sharing approaches directed at one or
management functionality (OAM). This information includes
the other, as shown in Fig. 11. The expected capacity needs of
carrier load levels, sector and site capabilities, information on
5G have led to considerations of introducing spectrum sharing
mobility, etc.
between mobile network operators. Licensed spectrum sharing
The repository and controller work together to collect
calls for a novel architectural solution that would respect the
spectrum usage information, distribute this information to
rights of licensees while ensuring quality of service. The
stakeholders, and control access to licensed frequency bands
goal is to dynamically allocate spectrum to licensed users
based on predetermined policy agreements. A live trial of
as it becomes available while protecting established license
the LSA/ASA concept was tested successfully on a 2.3 GHz
agreements.
band with existing network equipment, demonstrating that
dynamic availability of the LSA/ASA band can be handled
A. Licensed Spectrum Access (LSA) Concept with a minimal number of additional components, namely the
A new shared spectrum concept called Licensed Shared repository and controller [22].
Access has received considerable attention in European regu-
lation and standardization. According to the European Com- IV. T OWARD A U NIFIED 5G R ADIO ACCESS
mission, LSA is ”a regulatory approach aiming to facilitate A RCHITECTURE
the introduction of radio communication systems operated by The preceding analysis has reviewed architectures that have
a limited number of licensees under an individual licensing been proposed as independent systems, but these architectures
regime in a frequency band already assigned or expected are better viewed as complementary aspects of the 5G mobile
to be assigned to one or more incumbent users. Under the network. One key takeaway is that they all attempt to represent
Licensed Shared Access (LSA) approach, the additional users a heterogeneous network, one composed of sundry wireless
are authorised to use the spectrum (or part of the spectrum) devices with different underlying radio access technologies, at
in accordance with sharing rules included in their rights of an abstraction-layer high enough to centrally regulate, using
use of spectrum, thereby allowing all the authorized users, recent innovations from cloud computing and software-defined
including incumbents, to provide a certain QoS” [22][23]. This networking.
is a regulatory framework that demands a technical solution. The first step to establishing a cohesive 5th generation
LSA is an extension of Authorised Shared Access (ASA) radio access network is to decouple the control from the
to radio communications systems in general, whereas ASA is data plane, as RadioVisor, PRAN, and SDRAN attempt to
solely associated with mobile broadband. At its basic level, the do, by assigning responsibility to one or more centralized
system is intended to license out allocated but unused spectrum controllers. The second step is to virtualize network functions
to new users, respect incumbent users (if a tiered licensing by writing them in software, as PRAN does, using RadioVisor
system were implemented, for example, users with higher to manage physical radio resources. The third step requires
authorized priority should gain access over those with lower the introduction of an effective software-based management
status), and provide a satisfactory QoS. Apart from regulatory system, as SDRAN has done with its centralized control unit
considerations, a technical solution would have to maintain up- at the edge of the network. These technologies provide a
to-date records of relevant shared spectrum information, such comprehensive foundation for a heterogeneous RAN that can
as license-holders, incumbents, available frequency bands, and be managed as an abstract single system. On top of this system,
policies associated with each band, as well as control access the distributed spatial database model and LSA/ASA concept
to the shared spectrum. The concept can be implemented by can provide a far more flexible base station information service
introducing two new building blocks on top of the existing and spectrum sharing environment, respectively.
8
Fig. 10: Illustration of the software-defined radio access network (SDRAN). Figure by [15].
R EFERENCES
[1] E. Piri and H. Schulzrinne, “Scaling network information services to
support HetNets and dynamic spectrum access,” IEEE J. Commun.
Netw., vol. 16, no. 2, pp. 202–208, Apr. 2014.
[2] ITU-R, “Assessment of the global mobile broadband deployments and
forecasts for International Mobile Telecommunications,” Radiocommu-
nication Sector of ITU, Rep. ITU-R M.2243, 2011.
[3] Cisco, “Cisco visual networking index: Global mobile data traffic
forecast update, 2016-2021,” Cisco, White Paper, 2017.
[4] J. Rodriguez, Future of 5G Mobile Networks, 1st ed. Wiley, 2015.
[5] J. Andrews et al., “What Will 5G Be?” IEEE J. Sel. Areas Commun.,
vol. 32, no. 6, pp. 1065–1082, May 2014.
[6] ITU, “World Radiocommunication Conference allocates spectrum for
future innovation,” WRC-15 Press Release, 2015.
[7] Ericsson, “5G Radio Access,” Ericsson, White Paper, Uen 284 23-3204
Rev C, 2016.
[8] J. Kurose and K. Ross, “Computer Networks and the Internet,” in
Fig. 12: Key stakeholders and architecture for LSA/ASA. Computer Networks: A Top Down Approach, 7th ed. Hoboken, New
Figure by [22]. Jersey, USA: Pearson, 2017, ch. 1, p. 3.
[9] (2017, Feb.) RAT: Radio Access Technology. Technology Training Ltd.
[Online]. Available: http://technology-training.co.uk/rat.php
[10] B. Koley, “Software Defined Networking at Scale,” Google Technical
Infrastructure, BTE, 2014.
[11] M. Agiwal et al., “Next Generation 5G Wireless Networks: A Com-
prehensive Survey,” IEEE Commun. Surveys Tuts., vol. 18, no. 3, pp.
1617–1655, Third Quart. 2016.
[12] B. Rong et al., 5G Heterogeneous Networks, 1st ed. Springer, 2016.
[13] M. Fitch et al., “A white paper on spectrum sharing,” European
Commission, Radio Access and Spectrum (RAS) White Paper, 2012.
[14] I. Grigorik, “Mobile Networks,” in High Performance Browser Network-
ing, 1st ed. Sebastopol: O’Reilly, 2013, ch. 7, p. 110.
[15] S. Sun et al., “Heterogeneous Network: An Evolutionary Path to 5G,”
in Proc. 2015 21st APCC. IEEE, Oct. 2015, pp. 174–178.
[16] A. Gudipati et al., “RadioVisor: A Slicing Plane for Radio Access
Networks,” in Proc. 3rd ACM SIGCOMM Workshop Hot Topics Softw.
Def. Netw. - HotDSN. ACM, Aug. 2014, pp. 237–238.
[17] W. Wu et al., “PRAN: Programmable Radio Access Networks,” in Proc.
13th ACM Workshop Hot Topics Netw. - HotNets-XIII. ACM, Oct. 2014,
pp. 6–13.
[18] A. Gudipati et al., “SoftRAN: Software Defined Radio Access Network,”
in Proc. 2nd ACM SIGCOMM Workshop Hot Topics Softw. Def. Netw.
- HotDSN. ACM, Aug. 2013, pp. 25–30.
Fig. 13: A centrally controlled RAN architecture based on the [19] M. Rouse. (2017, Feb.) Long Term Evo-
technologies examined in this paper. lution (LTE). TechTarget. [Online]. Available:
http://searchmobilecomputing.techtarget.com/definition/Long-Term-
Evolution-LTE
[20] E. Haleplidis et al., “Software-Defined Networking (SDN): Layers and
technologies. Much research has been devoted to improving Architecture Terminology,” IETF, RFC 7426, Jan. 2015.
[21] P. Rysavy, “Challenges and Considerations in Defining Spectrum Effi-
the radio access and spectral efficiency of existing radio access ciency,” Proc. IEEE, vol. 102, no. 3, pp. 386–392, Mar. 2014.
networks. However, less attention has been paid to synthesiz- [22] M. Palola et al., “Live field trial of Licensed Shared Access (LSA)
ing existing research for the purposes of establishing a unified concept using LTE network in 2.3 GHz band,” in Int. Symp. Dynamic
Spect. Access Netw. (DYSPAN). IEEE, Apr. 2014, pp. 38–47.
5G RAN architecture. After examining innovations in both [23] “RSPG Opinion on Licensed Shared Access,” European Commission,
radio access and spectral efficiency, it is possible to discern Radio Spectrum Policy Group, RSPG13-538, Nov. 2013.
an emergent radio access infrastructure capable of meeting [24] T. chen et al., “Towards Software Defined 5G Radio Access Networks,”
in IEEE Softw. Def. Netw. Newslett. IEEE, Mar. 2016.
the RAN efficiency standards of 5G. Although a coherent [25] J. Gozalvez, “Tentative 3GPP Timeline for 5G [Mobile Radio],” IEEE
picture of the 5G RAN is beginning to develop, more research Veh. Technol. Mag., vol. 10, no. 3, pp. 12–18, Sept. 2015.
is needed into the challenges underlying each of the sub-
architectures as well as the interfacing technology required to
integrate them under a single service model. The tremendous
complexity of 5G networks will be much more tractable
10