Towards Software Defined 5G Radio Access Networks

Tao Chen, VTT Technical Research Centre of Finland; and Navid Nikaein, EURECOM



Network softwarization and virtualization are two key ingredients to abstract infrastructure resources and enable the delivery of the network as a service. Their tight coupling provides the flexibility needed to provision network resources on-demand and to compose and chain network service functions dynamically to meet a wide range of use-cases. Software Defined Network (SDN) and Network Function Virtualization (NFV) technologies are expected to play an important role in 5G networks. Both SDN and NFV have been widely applied in mobile core network design. In this article we present the concept of software defined Radio Access Networks (RAN). We believe software defined design in RAN will be a key step to support network slicing, RAN sharing, flexible spectrum management and other key features in 5G networks.


Software Defined Network (SDN) is seen as a promising technology to reshape the design of enterprise networks, metro networks, datacenters and Internet. For mobile networks, SDN and Network Function Virtualization (NFV) will play important roles to orchestrate mobile core networks. It is natural to apply SDN in mobile core networks as the primary functions of mobile core networks is about packet forwarding, along with other key functions including authentication, billing, and mobility management. The natural question is what the role of SDN is in RANs. Since SDN is proposed to manage the network complexity, the same principles of SDN may help to tackle different kinds of complexity in next generation heterogeneous RANs. In this article we will identify needs, address business impacts, and propose some initial thoughts on how to bring a software defined networking approach to designing, deploying and managing the RAN.

Needs for software defined RANs

There are lots of discussions on what 5G networks will be. While main technologies for 5G networks are yet to be specified, use cases and requirements have been well identified. NGMN and 3GPP have defined a bunch of 5G use cases [1][2]. In Europe the 5G PPP program funded by the Europe Commission has defined the key performance indicators of 5G networks [3]. The common understanding is that 5G will become a unified service platform to serve services covering mobile broadband, massive Internet of Things (IoT), and critical machine type of communications. Certainly new network architecture and technologies are required to fuel such a wide spectrum of services.

From the RAN perspective, the softwarization will facilitate not only the orchestration of radio resources but also the management and sharing of spectrum. The increasing complexity in 5G RANs calls for the new network design. New frequency bands and flexible spectrum access will be needed for the success of 5G. The software defined RAN could allow spectrum to be managed more efficiently, by spectrum awareness, spectrum mobility and effective implementation of spectrum sharing strategies in networks.

Moreover, SDN will be a key enabler to manage the complexity in 5G RAN. The complexity in the access layer (i.e. MAC and PHY) comes from the strong need to handle inter-cell radio resource management. The densified deployment of small cells in 5G network provides users rich accessibility but more complex interference scenarios. The access layer may heavily rely on network level coordination to function properly, to support coordinated multiple point transmission (CoMP) or network multi-input multi-output (MIMO), and more generally coordinated signal processing.

The complexity in heterogeneous RANs also comes from dynamic scheduling of wireless resources. Thanks to the flexible spectrum allocation in LTE, the joint frequency and time scheduling becomes a necessary approach to provision radio resources and to mitigate inter-cell interference. However, current distributed solutions adopted by LTE macro-cells are not scalable for dense small cell deployments. A SDN approach for RAN could offer flexible coordination schemes, in which the virtualization of radio resources at the network level could improve resource provisioning, node cooperation and mobility among the network.

5G RAN requires flexibility to support network slicing and service chaining on the fly, for a variety of services from different vertical sectors, e.g. automobile, health, energy and public safety. To improve the user experience and quality of service, RAN may heavily rely on the cloud computing and Mobile Edge Computing (MEC) design patterns for improved capability and capacity. Software defined implementation of RAN will be the key to realize programmability in the radio access layer.

Business impacts

The development of software defined mobile networks will have profound business impact on the value chain of mobile industry. Operators will be able to reduce the Capital Expenditure (CAPEX) and time to deploy services, because new open control interfaces and the software-defined control will reduce time and cost to reconfigure and optimize RANs, and to introduce new network features. The software defined control architecture will allow more efficient use of spectrum, energy resource, as well as the network infrastructure so as to reduce the Operational Expenditure (OPEX).

The open and flexible architecture will allow network equipment vendors the flexibility to implement network functions and services. It will reduce the time-to-market, and allow open innovation by embracing competition.

Content providers could have their Over-The-Top (OTT) services being better served. RANs can be tuned for OTT services by network slicing and RAN sharing. It provides content providers and Mobile Network Operators (MNO) a cooperation framework to benefit their business.

End users will receive smoother network experience by the improved coordination among different mobile networks. MNO is able to provide customized control to satisfy certain subscriber groups, and to deploy new services in shorter time, and further to create more value-added services.

Towards Software defined 5G RAN

Based on SDN design principles, one design example of software defined RAN is depicted in Figure.1.

With the necessary function extension, the network equipment is able to report status information of cells to an introduced control layer, which could be implemented by, e.g., MEC techniques. The control layer has also the repository to store policies and spectrum information of the network. The regional view of cells, according to the correlation of the cells, is generated by extracting the information from the repository. This view includes spectrum allocation, connection of user devices, and interference map among cells.

With network virtualization functions, the regional view of cells is further abstracted to the high level abstracted view, which is used by control applications. In addition, the history of configuration and statistics of user data stored in the repository can be processed or learnt for the better configuration of the network.

On top of the control framework there are programmable control applications for different control purposes. When a control is made by control applications, it is mapped to control commands at the device level from the network virtualization layer.

Figure 1

Figure.1 Software defined control framework for heterogeneous RAN

The time-critical control can be supported through low level control automation or delegation to underlying RAN modules. These modules take charge of coordination and cooperation among cells, interpret the requirements and set up the signaling path for the cooperation nodes. It receives control instructions from the high level of the control plane and responses to set up control signaling paths. Depending on the requirements on delay, signaling volume, and control method, the path may be over-the-air or from backhauls of cells. After the setup of the control paths, the involved nodes use signaling to configure properly for the cooperation. So the aim is to provide an open platform for real-time inter-cell cooperation, not limited to specific cooperation algorithms.

Figure.2 illustrates the network slicing supported by the software defined design of mobile networks. The lowest sub-layer in the figure includes processing, storage, wired/wireless network elements, as well as, Radio Frequency (RF) front-ends. The virtual resources (i.e., cloud regions) that are controlled by the infrastructure manager following ETSI NFV [4] and/or SDN methodologies and mechanisms reside on the upper second sub-layer. The virtualization layer is a key enabler for Infrastructure as a Service (IaaS) clouds. Every slice's virtual infrastructure must be isolated from other slices to operate efficiently without violating the special performance requirements (performance isolation). Consistency Availability and Partition tolerance (CAP) conjecture states that it is impossible for any distributed system to provide consistency, availability, and partition tolerance at the same time. Thus, a trade-off must be made at the design time for different slice regions to control the partitioning and fulfil two out of the three guaranties.

Figure 2

Figure.2 Network slicing in software defined mobile networks

In order to maximize the infrastructure utilization, it is required to dynamically and freely relocate hardware resources depending on current and local needs, under the control of cloud operators (be full or partial). Virtual resources, virtual services, service bundle, and service chains, cloud as well as NFV and SDN technologies must be leveraged towards rapid building. NFV enables for extreme flexibility when it comes to the micro-service architecture, service chaining, and service life cycle management, hiding all the configuration complexities of underlay resources (physical and virtual). As an example, different network functions and/application components have special requirements in terms of processing time, delay, and jitter. They require to be adequately provisioned on appropriate platform configurations that can accommodate for the intended application requirements.

Regarding the network segment for every slice, the application layer of the architecture has to remain agnostic of the real network topology that is used by the virtual network provider, while the network slice operator must be able to determine the exact status of the network resources. This requires identifying (a) how network state and resources are exposed (and represented) to enable network application and services, (b) how the (virtual) network functions are aggregated and composed to construct the network, and (c) how the network is configured and instantiated for a particular application domain. SDN is emerging as a natural solution for next generation cellular networks, as it enables further network function virtualization opportunities and network programmability. We plan to leverage SDN technologies and use SDN controllers together with programmable data plane technologies in order to support advanced programmability of the mobile networks (switching) over radio data-plane abstractions.


This article provides a brief introduction on software defined RANs. For 5G it is not only about new spectrum and new air interface, but also about the new architecture and network technologies. The exploitation of cloud technologies, SDN, NFV, and MEC can provide the necessary tools to break-down the current vertical RAN design into a set of horizontal mirco-service network functions. It has the potential to change the ecosystem of mobile networks, to enable truly mobile Internet. Under the 5G PPP research program, the COHERENT project is dedicated to this aspect. For more information, please visit


[1] NGMN Alliance, “5G White Paper,” Next Generation Mobile Networks, White paper (2015).

[2] 3GPP TR 22.891 Draft, “Feasibility Study on New Services and Markets Technology Enablers, Stage 1 (Release 14)”, v1.1.0 (2015-11), Nov. 2015

[3]5G PPP, “5G Vision: The 5G Infrastructure Public Private Partnership: The next generation of communication networks and services,” 5G PPP, Feb. 2015

[4] ETSI, “Network Functions Virtualization (NFV) Architectural Framework (2014),” available at:



Tao ChenDr. Tao Chen is a senior researcher at VTT Technical Research Centre of Finland. He received his Ph.D. degree from University of Trento, Italy in 2007 and B.E. degree from Beijing University of Posts and Telecommunications in 1996. From 2003 to 2007, he worked with CREATE-NET, Italy. Since 2008 he has been with VTT, working on cognitive wireless networks, green communications and 5G network technologies. Currently he is the project coordinator of Europe H2020 5G PPP COHERENT project (, which is one of major 5G research projects in Europe. He is the board member of EU 5G PPP Steering Board ( He is an IEEE senior member. His research interests include: dynamic spectrum access, energy efficiency and resource management in heterogeneous wireless networks, software defined networking for 5G mobile networks, and social-aware mobile networks.


Navid NikaeinDr. Navid Nikaein is an assistant professor in mobile communication department at Eurecom. He received his Ph.D. degree (docteur es sciences) in communication systems from the Swiss Federal Institute of Technology, EPFL, in 2003. He is leading a research group focusing on experimental system research related to wireless systems and networking. His contributions are in the areas of wireless access layer techniques and networking protocols, cloud-native and programmable radio networks, and real-time prototypes and scalable emulation and simulation. He has a proven track record in managing both fundamental and experimental research projects at both European and an international level and is leading the development of the radio access layer of OpenAirInterface software platform.



Laurent CiavagliaLaurent Ciavaglia is currently senior research manager at Nokia Bell Labs where he coordinates a team specialized in autonomic and distributed systems management, inventing future network management solutions based on artificial intelligence.

In recent years, Laurent led the European research project UNIVERSELF ( developing a unified management framework for autonomic network functions. , has worked on the design, specification and evaluation of carrier-grade networks including several European research projects dealing with network control and management.

As part of his activities in standardization, Laurent participates in several working groups of the IETF OPS area and is co-chair of the Network Management Research Group (NRMG) of the IRTF, member of the Internet Research Steering Group (IRSG). Previously, Laurent was also vice-chair of the ETSI Industry Specification Group on Autonomics for Future Internet (AFI), working on the definition of standards for self-managing networks.

Laurent has co-authored more than 80 publications and holds 35 patents in the field of communication systems. Laurent also acts as member of the technical committee of several IEEE, ACM and IFIP conferences and workshops, and as reviewers of referenced international journals, and magazines.



Subscribe to IEEE Softwarization

Join our free SDN Technical Community and receive IEEE Softwarization.

Subscribe Now


Article Contributions Welcomed

Download IEEE Softwarization Editorial Guidelines for Authors (PDF, 122 KB)

If you wish to have an article considered for publication, please contact the Managing Editor at


Past Issues

November 2018

March 2018

January 2018

December 2017

September 2017

July 2017

May 2017

March 2017

January 2017

November 2016

September 2016

July 2016

May 2016

March 2016

January 2016

November 2015

IEEE Softwarization Editorial Board

Laurent Ciavaglia, Editor-in-Chief
Mohamed Faten Zhani, Managing Editor
TBD, Deputy Managing Editor
Syed Hassan Ahmed
Dr. J. Amudhavel
Francesco Benedetto
Korhan Cengiz
Noel Crespi
Neil Davies
Eliezer Dekel
Eileen Healy
Chris Hrivnak
Atta ur Rehman Khan
Marie-Paule Odini
Shashikant Patil
Kostas Pentikousis
Luca Prete
Muhammad Maaz Rehan
Mubashir Rehmani
Stefano Salsano
Elio Salvadori
Nadir Shah
Alexandros Stavdas
Jose Verger