1
SDN & NFV in 5G: Advancements and Challenges
Christos Bouras∗† , Anastasia Kollia† , Andreas Papazois†
∗ Computer Technology Institute & Press ”Diophantus”, Patras, Greece
† Computer Engineering & Informatics Dept., University of Patras, Greece
Abstract—The next generation of mobile networks has risen
the expectations and has induced a whole new approach in the
networking future. This study summarizes solutions, that meet all
the existing requirements integrating Software Defined Network
(SDN) and Network Function Virtualization (NFV) in mobile
networking. The main contribution of this research activity is that
it is a tool for literature reference for scientists and engineers,
because it presents and compares the most fundamental existing
solutions and the future directions. In this paper, the main
characteristics of SDN and NFV are summarized as fundamental
alternatives contributing to succeed the goals set by 5G. The
main demands and challenges faced by these solutions and the
most beneficial applications of the combination of SDN with NFV
are described. Finally, the main conclusions of this research and
suggestions for possible future activity in the field are described.
Fig. 1: The general concept of SDN architectural model.
Keywords—SDN, NFV, 5G, mobile networks
The remaining part of this paper is structured as follows: in
I. I NTRODUCTION Section II the main challenges SDN and NFV should meet are
The arrival of the next generation of mobile networks has listed. In Section III the SDN mobile and wireless architectures
rendered the demands of novel, more evolved and scalable and the uses of NFV in mobile networks are described. In
technologies an imperative requirement. Mobile users are Section IV there is a comparison and contrast of the suggested
highly augmenting not only because of the fact that the mobile solutions. In Section V the expected evolution in the
number of personal devices is incrementing, but also because domain in the next years is outlined, the paper is concluded
of (machine to machine (M2M) communication methods, the and some ideas for future research activity are listed.
Internet of Things (IoT)), that add data overhead, augment the
data rate, raise the capacity demands and increase the need for II. R EQUIREMENTS & C HALLENGES
coverage. In this section, the main requirements and obstacles that
Although several types of mobile networks will dominate in have been raised are summarized.
the 2020’s, there are also many challenges, such as: extenu- Fig. 1 describes the concept in the already presented SDN
ating power consumption in devices and base stations, better architectures, such as [12] and [16], which are based on
resource allocation, higher data rates, ensuring lower round trip decoupling the data and control layers. The infrastructure layer
times, reducing all costs, optimization of mobility management includes commodity switches. The control layer includes a
policies, scalability, elasticity and agility. cluster of controllers. The application layer includes applica-
Though until recently decentralized architectures were con- tions and network appliances. Firewalls are virtually deployed
sidered safe and efficient leading to more and more decen- reclaiming the NFV and function in the application layer.
tralized models, the augmenting network traffic advocates in The most vital requirements that are linked to the SDN
favor of adopting a centralized way of control. A central archi- wireless and mobile cases and the demands of NFV presented
tecture provides important benefits, such as optimal frequency [6] are summarized in Table I. It is significant to standardize
assignment and efficient mobility management policies. the control and provision future demands of resources and
Software Defined Networking (SDN) is not only a means of succeed in providing dynamic control not only by statistical
meeting the 5G demanding goals, but also offers an alternative development and real-time data. Viewing a large part of the
solution to face the main mobile network problems. network, contributes in enhancing the assignment and routing
There is not much activity in the field of reviewing the capabilities.
most important studies including SDN and NFV, although It also provides performance degradation control policies,
there is much research in possible applicable scenarios for requirements in safety and security issues and alternative
implementing the technologies. The survey [16] summarizes applicable scenarios in case the controllers get out of order. 5G
the most important facts about SDN and includes several other and beyond networks are based on ultra-dense deployments.
studies and testings in the field. A framework for cellular SDN It is important to succeed in providing denser cellular
is presented thoroughly in [7]. The main objective of this paper infrastructures, increasing the available bandwidth, exploring
is to summarize the main features of SDN and NFV. Multiple Input Multiple Output (MIMO) solutions collecting
The survey [16] summarizes the most important facts about network state, specifying the behavior of the network, updating
SDN and includes several other studies and testings in the field. the network configuration. The OpenFlow protocol and the
Open Network Operating System (ONOS) is the most well- wireless networking abstractions contribute to this direction,
known and important SDN controller. Its latest releases (from namely the light virtual access point, resource pool, interfer-
Emu and beyond) introduce the Central Office Re-imagined ence map [20]. Finally, the authors end up evaluating the
as a Data center (CORD). The next release of NFV is the suggested resource management, querying and interference
Brahmaputra, is going to be combined with these releases. tracking.
TABLE I: The main requirements and challenges linked to the SDN and NFV.
SDN NFV
CONTROL
1. Standardization of the control interfaces 1. Seamless control and provisioning
2. Protection of commercial business operating schemes 2. Real time and dynamic provisioning
3. Measures to avoid performance degradation 3. Creation of network granularity policies
4. Maintenance of information of the controlling network-big 4. Maintenance of virtualization information-big data
data development development
RELIABILITY
1. Seamless connectivity and fast connection recovery 1. High complexity of 5G (technologies, devices, IoT)
2. Security requirements in EPC and RAN 2. Seamless and high quality connectivity
3. Security and reliability of the transport and data network 3. Virtualization of terminal points
4. Equilibrium among performance, security and flexibility 4. Security concerns (same physical medium)
SCALABILITY
1. Support of technology and device heterogeneity 1. Carrier-grade scalability and robustness
2. Controller messages with performance and survivability (low 2. Acceleration of implementation
packet loss levels)
3. Optimization of flow rules-better network slicing 3. Openness and interoperability, global reach and cross-
administration
COST EFFICIENCY
1. Capability to support a pay for service commercial model 1. Power consumption reduction
2. Replacement of hardware with software applications 2. Operational efficiency improvement
3. Deployment and acquisition of commodity switches- 3. Higher capital costs
Replacement of legacy hardware
4. Shorter time to market and less deployment risks 4. Higher operational costs (short life-cycle of configu-
ration tools)
There is an introduction of some basic terms for IP mo- solutions, such as mobile or cellular SDN to succeed in
bile networks. They analyze the procedure of exchanging IP the optimization of the mobile networking. Analytically, the
datagrams. There is also a summary of the main mobile IP Mobility Management Entity (MME), the Serving Gateway
challenges, such as security, triangle routing and handover. It (S-GW), the Packet Data Node Gateway (P-GW), the Policy
describes the OpenFlow implementation based on registration and Charging Rules Function (PCRF), the Home Subscriber
processes, packet processing and the controller application. It Server (HSS), the cloud RAN (OpenRadio, OpenRAN, etc),
concludes that this research activity enforces seamless net- the Remote Radio Units (RRU), the Virtualized Base Station
working mobility and improves QoE [19]. - Baseband Units (BBU) are presented. The control and the
Review [18] describes a base station and a core network data planes are split. The mobile version of SDN includes all
virtualization. The most important motivations of using the the components existing into the cloud figure. It also manages
NFVs are analyzed. For NFV cases, the virtualized and not Quality of Service (QoS) and provides Deep Packet Inspection
virtualized mobile core network should coexist according to (DPI). Finally, Policy and Charging Rules Function (PCRF)
[18]. supports service data flow detection, policy enforcement and
The challenges and performance degradation effects are flow-based charging.
avoided in the network by the SDN and are specially controlled
by the ONOS project in the mobile 5G [5].
B. Existing Solutions
1) OpenRadio: [4] describes the programmable data plane
III. P ROPOSED SOLUTIONS OF M OBILE SDN/NFV
in wireless systems and its possible implementation in today’s
In this section, the main aspects of the existing solutions in mobile networks. Firstly, there are several analyses including
the mobile SDN cases are presented. the cell-based optimization, the coexistence of heterogeneous
and alternative types of cells and the application of specific
A. Overview wireless services. It introduces programming abstractions, such
as modular declarative interface, information plane, state ma-
The EPC in SDN solutions is virtually deployed. The NFV chine model and deadlines. Finally, it refers to the challenges
concept will be used for the case of EPC so that the network of the OpenRadio and suggests a designing policy.
operates in a virtualized way. The idea takes advantage of all 2) OpenRAN: [21] introduces an OpenRAN approach
NFV assets including scalability, lower costs etc. According via virtualization. The suggested overall protocol is divided
to [12] and [14] EPC provides substantial applications: in the SDN controller, in the wireless spectrum resource
• Implementation of virtualized mobile gateways, S-GW, and the cloud computing resource protocol. It also induces
P-GW. four levels of virtualization: application, cloud, spectrum and
• Creation of mobility management policy. cooperation. It concludes that controllers and programmable
• Network management & network components manage- schemes achieve routing and bandwidth allocation and several
ment. flow priorities.
• Band splitting in cloud cases & management of cloud 3) SoftRAN: [9] introduces the dense network manage-
nodes. ment. The authors design the SoftRAN (Software defined
In Fig. 2 the overall SDN mobile architecture based on RAN) combined with a coupled control plane in dense net-
the today’s LTE networks is summarized. It describes the works. It introduces the base station abstraction, the SoftRAN,
RAN, the EPC components, the virtualized network parts the controller architecture and the refactoring of the control
and how it is possible combining these with mobile SDN plane. It ends up indicating the feasibility of RAN.
Fig. 2: The overall architectural scheme, including the EPC, RAN, mobile SDN and NFV.
4) NetShare: [17] operates as manager of shared resources 8) Wireless SDN (WSDN): [6] introduces a wireless SDN
and is introduced as a way of reallocating the existing re- architecture. The virtualization of several functions and mech-
sources of base stations and multiple entities in an existing anisms of the network infrastructure are analyzed alongside
network. Netshare isolates entities and highly utilizes the with other important issues, such as sharing, virtualizing core
resources of an entity. The resource allocation is based on servers, dynamic energy policies. SDN, OpenFlow, CAPWAP
distribution, strict isolation, network heterogeneity and per and reconfigurable wireless devices are defined. Virtualization
base station reservation. matters, QoE-aware network operations, network access selec-
5) ONOS: [5] analyzes the SDN based on ONOS and the tion and mobility control are raised.
motivation linked to wireless transport networks. MCORD 9) SDN Mobile Cloud: [11] offers important benefits in
(Mobile CORD) is the mobile ONOS CORD and is available wireless communications alongside with novel applications,
for experimentation since the Emu release of ONOS. that stem from SDN. There is a consideration of several
6) Software Defined Cellular Network (SDCN): [14] challenges of combining SDN with adhoc networks and an
includes fundamental answers in the already augmenting heterogeneous architecture is proposed.
demands, namely flexible policies, scalability, commodity 10)Mobile Extension of SDN (MeSDN): [12] summarizes
switches, remote control of base stations. Several controller the possible mobile cloud requirements.MeSDN should be
applications are directing traffic throughout middleboxes, mon- used at the enterprises to cover the augmenting demands of
itoring network controlling and billing, providing seamless the visitors in using the network. The proposed architecture
subscriber mobility and QoS, accessing control policies, of- describes the flow manager, the roles of the local and the global
fering virtual cellular operators and managing intercell inter- controllers. There is also a reference in MeSDN applications,
ference. Last but not least, slicing several resources, such as such as App-aware End to End (E2E), QoS, network fault
bandwidth, topology, traffic, device CPU and forwarding tables diagnosis, WLAN Virtualization, 4G cellular networks. It
is important for efficient resource allocation. introduces pTDMA (pseudo Time Division Multiple Access),
7) Cellular SDN (CSDN): [15] focuses on flexible policies a type of pseudo TDMA, analyzing scheduling principles,
on subscriber attributes, scalability through local agents, flex- downlink control, power saving, prototype implementation,
ible switch patterns and actions and network virtualization on such as architecture, threats, evaluation. The most important
subscriber attributes. There are also several CSDN applications challenges are presented, among which are: millisecond level
such as flexible subscriber policies, scalability for local agents, synchronization, driver buffering delay and 802.11 beacons.
flexible switch patterns and virtualization of the subscriber
attributes. C. Ultra-dense deployments based on SDN
The SDN architecture induces logical centralized control, A network model and some control functions are presented,
programmability and high abstraction levels [10]. The back- in which the controllers are divided in two possible categories
ground study is based on inter-cell interference management based on optimization approaches. For long term optimization,
and mobile traffic management. the technique is deployed in dedicated hardware, while for
SDN and NFV are capable of providing intelligent services short term ones only cover some base stations [2]. There
and dynamic resource allocation. The virtualized network in is a presentation of some controller applications, such as
SDN architectures includes several functions, designing con- mitigation LTE, WLAN optimization, LTE access selection,
siderations in the forwarding and control plane [7]. Practical power cycling and offloading.
questions are considered when it comes to centralized vs The main security dangers in 5G, such as interference,
distributed data collection, data processing, decision making backhauling, inconsistent security and high complexity, due
and user privacy. Future work is summarized as being a to augmenting number of devices, are described thoroughly.
combination of CSDN with Big Data. The exact applicability of SDN in mobile networks is analyzed
alongside with the fact that constant handovers raise security hand, the solution helps reducing the CAPEX and OPEX for
issues in 5G. In order to avoid these problems 5G is suggested a provider and the time for implementing a new service.
to be implemented using SDN capabilities, handover mecha- Table II summarizes the basic comparison of the most
nisms and privacy protection [8]. fundamental features of different SDN suggestions.
IV. C OMPARISON OF SDN & NFV SOLUTIONS V. C ONCLUSIONS & F UTURE D IRECTIONS
In this section, there is a comparison between the two
In this section, possible future directions in the domain
main technological deployments. The most significant studies
of SDN and the particular suggestion are explained and are
combine the two achievements and contribute in outlining the
formed in future research activity.
future directions in the field.
The most important requirements of 5G delineate the future
• Motivation of deployment: SDN is deployed in order directions in the field of mobile SDN, most of which are
to decouple the control and the data plane and provides presented in Table III. The virtualization of networking nodes
network controlling and programmability. NFV is an and the cloud service applications offer high network perfor-
abstraction of network functions. mance (network speed, data rates, low packet loss, low round-
• Abstraction: Abstraction levels provided by the NFVs trip times, low power consumption etc) and lead to better
should be standardized. Radio Resource Management (RRM). The network is possibly
• Types of virtualization: It is possible to virtualize intelligent due to information acquired by the network traffic. It
resources or hardware devices, build virtual base stations is also fundamental to create particular policies and prototypes
and manage resources in mobile networking. for the controllers. The heterogeneous network deployments
• Mobility Management: The virtualization methods offer E2E solutions and provide QoE and QoS.
contribute to better mobility management, creation of In 5G it will be vital to design switches and controllers, that
policies and deployment of a billing system.It is valuable are compatible with the SDN technology or modify the existing
for controlling traffic classes and forming several flow ones with SDN enabled. Scalability is a very important issue
space partitioning rules. as the heterogeneous networks described in 5G are ultra-dense
• Heterogeneity: Heterogeneous solutions are supported, and their exact size is not known yet.
supporting combination of different technologies and The high speed rates and the low round trip times ensure
types of cells. that the network performance in 5G and beyond will be
• Efficient handover policies: It is important to standard- the best possible. So RRM is important to be investigated
ize innovative radio resource management functionali- and implemented in the future. The issue also raised is how
ties. resources are allocated to the applications and the development
• Cost efficiency and software based implementation: of a certain policy made in the SDN controller.
Network Address Translation (NAT), firewalls and mo- The need for supporting QoS also raises the importance of
bile core functions should be virtually deployed. describing and distinguishing its different classes. An impor-
• Open Source: Some of the open source solutions are tant issue is also to reclaim the information from the network in
going to be deployed as commercial ones providing extra a statistic and dynamic way in order to enhance coverage and
features in their commercial version. other resource demands. Finally, E2E SDN solutions should
• Location: SDN controllers are located at the data center, be suggested and implemented as a way to meet the demands
whereas NFVs are located in service provider networks. of 5G and beyond.
• Prototypes: SDN is prototyped for the wired case and The combination of SDN and NFV promise a different
there are several suggestions for the mobile and wireless concept of networking deployments as long as the main
cases, there is not a prototype for NFV. challenges presented are faced.
• Technology Readiness Level (TRL): is a method rank- In general this research summarizes the existing SDN and
ing the maturity levels of a technology. It scales from 1 NFV solutions for the mobile case with an insight in 5G
to 9 and represents the existence of basic technological networks. This paper is able to inform scientists of the latest
research in the field and the system launching and oper- trends in the domain and also consists a very strong tool, as
ation respectively. In our case, some of the suggestions it reviews several important studies in this particular domain.
are close to implementation, while others are in a more In the future, an architecture combining the SDN and NFV
immature level. is going to be presented as a new future activity and how based
• Applicability: Applicable scenarios will provide scal- on both a more optimized mobile network architecture will be
able and reliable solutions and will cover most of the succeeded and will cover the demands and needs of future
mobile network demands. It is also vital to prove their mobile users. The primary goal of the authors is to test and
applicability through thorough testing and evaluation evaluate a possible SDN architecture.
procedures.
SDN is located at the data center, whereas NFVs are located
in service provider networks. SDN applications are linked to R EFERENCES
cloud orchestration and network management, on the other [1] S. Ali, V. Sivaraman, A. Radford, and S. Jha. A survey of securing
hand NFVs are mostly linked to particular network compo- networks using software defined networking. Reliability, IEEE Trans-
nents, such as firewalls, gateways, content delivery network actions on, 64(3):1086–1097, Sept 2015.
(cdn). Although SDN and NFV are two extremely different [2] H. Ali-Ahmad, C. Cicconetti, A. de la Oliva, V. Mancuso, M. R. Sama,
technological suggestions, their combination offers benefits in P. Seite, and S. Shanmugalingam. An sdn-based network architecture for
favor of succeeding high network efficiency and performance. extremely dense wireless networks. In IEEE Software Defined Networks
for Future Networks and Services (IEEE SDN4FNS). IEEE, 2013.
As referred to [7] all radio resource management func-
tionalities are included. The corresponding virtual Evolved [3] M. Banikazemi, D. Olshefski, A. Shaikh, J. Tracey, and G. Wang.
Meridian: an sdn platform for cloud network services. IEEE Com-
Packet Core (EPC) functionalities are thoroughly described. munications Magazine, 51(2):120–127, 2013.
In [6] virtualization methods are linked to implementation [4] M. Bansal, J. Mehlman, S. Katti, and P. Levis. Openradio: a pro-
of different overlay networks. They are responsible for NAT grammable wireless dataplane. In Proceedings of the first workshop on
firewalls and implement cellular core functions. On the other Hot topics in software defined networks, pages 109–114. ACM, 2012.
TABLE II: Comparison of mobile SDN/NFV solutions.
Solution
MeSDN CSDN WSDN NetShare SoftRAN OpenRAN MCORD
Feature
Network as a Ser-
vice (NaaS)
NFV abstractions X X X X X
Heterogeneity X X X X X
Service provider X X X
Network
Scalability X X X X X X X
Reliability X X
Implementation X X
Evaluation-Testing X X X X X X
TRL 6 4 4 6 6 6 7
Network part EPC & RAN EPC & RAN EPC & RAN RAN RAN RAN EPC & RAN
TABLE III: 5G requirements, that should be meet by the mobile SDN combined with NFV.
Demand
Capacity High data rate Reliability Energy efficiency Mobility High coverage
Direction
Virtualization X X X X
Scalability of the network X X X
components
High performance X X X X X
Interface standards X
Cloud services/apps X X X X X
QoS & QoE X X X X X X
Controller prototyping X X
Efficient RRM X X X X X
E2E solutions X X
Software defined RAN X X X
[5] D. Bercovich, L. M. Contreras, Y. Haddad, A. Adam, and C. J. [16] Y. Liu, A. Y. Ding, S. Tarkoma, et al. Software-defined networking in
Bernardos. Software-defined wireless transport networks for flexible mobile access networks. 2013.
mobile backhaul in 5g systems. Mobile Networks and Applications, [17] R. Mahindra, A. Khojastepour, H. Zhang, and S. Rangarajan. Network-
20(6):793–801, 2015. wide radio access network sharing in cellular networks.
[6] C. Bernardos, A. De La Oliva, P. Serrano, A. Banchs, L. Contreras, [18] OPEN NETWORKING FOUNDATION. Network functions virtuali-
H. Jin, and J. Zuniga. An architecture for software defined wireless sation (nfv);architectural framework. Technical report, ETSI, October
networking. Wireless Communications, IEEE, 21(3):52–61, June 2014. 2013.
[7] A. Bradai, K. Singh, T. Ahmed, and T. Rasheed. Cellular software [19] P. Pupatwibul, A. Banjar, A. Sabbagh, R. Braun, et al. Developing
defined networking: A framework. Communications Magazine, IEEE, an application based on openflow to enhance mobile ip networks. In
53(6):36–43, 2015. Local Computer Networks Workshops (LCN Workshops), 2013 IEEE
[8] X. Duan and X. Wang. Authentication handover and privacy protection 38th Conference on, pages 936–940. IEEE, 2013.
in 5g hetnets using software-defined networking. Communications [20] R. Riggio, K. M. Gomez, T. Rasheed, J. Schulz-Zander, S. Kuklinski,
Magazine, IEEE, 53(4):28–35, April 2015. and M. K. Marina. Programming software-defined wireless networks.
In Network and Service Management (CNSM), 2014 10th International
[9] A. Gudipati, D. Perry, L. E. Li, and S. Katti. Softran: Software defined Conference on, pages 118–126. IEEE, 2014.
radio access network. In Proceedings of the second ACM SIGCOMM
workshop on Hot topics in software defined networking, pages 25–30. [21] M. Yang, Y. Li, D. Jin, L. Su, S. Ma, and L. Zeng. Openran: a software-
ACM, 2013. defined ran architecture via virtualization. In ACM SIGCOMM computer
communication review, volume 43, pages 549–550. ACM, 2013.
[10] M. H. Kabir. A novel architecture for sdn-based cellular network.
International Journal of Wireless & Mobile Networks, 6(6):71, 2014. [22] K.-K. Yap, R. Sherwood, M. Kobayashi, T.-Y. Huang, M. Chan,
N. Handigol, N. McKeown, and G. Parulkar. Blueprint for introducing
[11] I. Ku, Y. Lu, and M. Gerla. Software-defined mobile cloud: Architec- innovation into wireless mobile networks. In Proceedings of the Second
ture, services and use cases. In Wireless Communications and Mobile ACM SIGCOMM Workshop on Virtualized Infrastructure Systems and
Computing Conference (IWCMC), 2014 International, pages 1–6. IEEE, Architectures, VISA ’10, pages 25–32, New York, NY, USA, 2010.
2014. ACM.
[12] J. Lee, M. Uddin, J. Tourrilhes, S. Sen, S. Banerjee, M. Arndt, K.-H.
Kim, and T. Nadeem. mesdn: Mobile extension of sdn. In Proceedings
of the Fifth International Workshop on Mobile Cloud Computing &
Services, MCS ’14, pages 7–14, New York, NY, USA, 2014. ACM.
[13] L. Li, Z. Mao, and J. Rexford. Toward software-defined cellular
networks. In Software Defined Networking (EWSDN), 2012 European
Workshop on, pages 7–12, Oct 2012.
[14] L. E. Li, Z. M. Mao, and J. Rexford. Toward software-defined cellular
networks. In Software Defined Networking (EWSDN), 2012 European
Workshop on, pages 7–12. IEEE, 2012.
[15] X. J. L. E. Li, L. Vanbever, and J. Rexford. Cellsdn: Software-defined
cellular core networks.
