Content uploaded by Angelos Antonopoulos
Author content
All content in this area was uploaded by Angelos Antonopoulos on Aug 23, 2018
Content may be subject to copyright.
A C-RAN based 5G platform with a fully virtualized, SDN
controlled optical/wireless Fronthaul
Kostas Ramantas, Angelos Antonopoulos*, Elli Kartsakli, Prodromos-Vasileios Mekikis, John Vardakas,
Christos Verikoukis*
Iquadrat Informatica S.L., Barcelona, Spain
Tel: (+34) 934678178, e-mail: kramantas@iquadrat.com
* Telecommunications Technological Center of Catalonia (CTTC/CERCA), Castelldefels, Spain
ABSTRACT
This paper describes in detail the design of a 5G platform based on the C-RAN architecture, with a fully
virtualized Radio Access Network (RAN) and an optical/wireless Fronthaul. The wireless and optical domains
are controlled by a hierarchy of Software Defined Networking (SDN) controllers, which are responsible for end-
to-end optimization of the platform, including the Fronthaul and 5G air interfaces. The proposed architecture
adopts a modular eNodeB (eNB) design, where virtualized BBU and RRH entities are implemented with
Commercial Off-Τhe-Shelf (COTS) components. Moreover, a mmWave RAU is connected to the BBU via a
feeder optical fiber, interconnecting one or more RRH units with the BBU over mmWave interfaces. COTS
components and Ethernet interfaces are employed for C-RAN prototyping of our platform, facilitating flexibility,
cost reduction and increased scalability.
Keywords: 5G testbed, C-RAN, vRAN, Fronthaul, SDN, NFV.
1. INTRODUCTION
Software Defined Networking (SDN) and Network Function Virtualization (NFV) are two promising
technologies that are expected to increase the efficiency of 5G networks, and enhance the flexibility of network
configuration and management [1]. In SDN, a centralized SDN controller, which handles network management
operations, is physically decoupled from the data plane and enables network programmability. Applications and
services running on top of 5G networks will need to be resource- and network-aware, in order to take full
advantage of the underlying network programmability, communicating with the SDN layer via a set of
Southbound Application Programming Interfaces (APIs) in a way that optimizes the resource allocation and
utilization in a centralized way. While SDN is typically associated with wired networks and data centers, several
advantages can be realized by SDN at the wireless domain as well. In the wireless domain, SDN enables the
implementation of centralized handover and load balancing schemes [2], thus optimally managing the scarce
wireless spectrum.
NFV is another cornerstone technology of 5G, which is employed to build an agile and programmable
virtualized infrastructure. NFV underpins the delivery, deployment, provisioning, and monitoring of services,
while Management and Orchestration (MANO) Frameworks that follow the ETSI NFV reference architecture
allow the infrastructure to adapt to service requirements, enabling fast and cost-efficient service management [3].
The next step in infrastructure virtualization is virtualizing the 5G RAN. This approach, which is known as
Virtualized RAN (vRAN), is recognized as a very promising area of innovation in the 5G ecosystem, resulting in
cost reductions and scalability benefits for 5G deployments. Specifically, it allows developing a C-RAN based
5G architecture with low-cost Whitebox servers and COTS components [4].
RAU
Core Tier
Whitebox
vBBU
vBBU
vBBU
Grid 1
Controller
vSGW
vPGW
vMME
VNFs
MANO
Apps
SDN Switch Fabric
SDN/NFV
RRH RRH
mmWave
fronthaul
MANO
Orchestrator
SDNi
Figure 1: Testbed Architecture
Several large-scale and academic 5G testbeds are currently being developed, allowing researchers to run their
own experimental protocols and evaluate future 5G architectures [5]. Small-scale platforms have also been
proposed using COTS components that reduce development costs [6]. Other works address the RAN part of 5G
networks, focusing on RAN Virtualization or flexible resource allocation [7], [8]. However, these are largely
conceptual works and do not consider the associated implementation challenges of real-world RANs. On the
other hand, the OpenAirInterface (OAI) software alliance, or OSA, implements an end-to-end 5G testbed based
on vRAN [9], which is freely available in the form of open source software. While not currently a production-
ready solution, its open source nature and flexibility allows a fast pace of innovation, as community members
can contribute and new developments are instantly available for experimentation. Moreover, a recent work in
[10] also moves past the theoretical context and focuses on the prototyping challenges of applying SDN
principles to solve the Radio Resource Management problem in 5G networks for a number of use cases.
In this paper, we propose a novel architecture for a C-RAN based 5G platform with a fully virtualized, SDN
controlled optical/wireless Fronthaul. Our testbed leverages state-of-the-art components from the OpenDaylight
([11]) and OpenAirInterface (OAI) open source projects, and employs the FlexRAN controller ([10]) for
centralized SDN control of the RAN. The proposed setup aims to enable the experimentation and validation of
novel algorithms and mechanisms in next generation 5G networks.
2. 5G testbed Architecture
The C-RAN based testbed architecture, depicted in Fig. 1, considers the application of the SDN and NFV
paradigms to fully virtualize the Core and RAN parts of a 5G network. This approach will allow us to employ
COTS components to implement all testbed tiers, significantly decreasing cost and increasing flexibility.
Employing a fully programmable, virtualized RAN (vRAN) on top of Whitebox servers and embracing Ethernet
and 802.11 interfaces allows a very cost effective C-RAN implementation, which was until recently only
available in "black box" carrier-grade equipment.
2.1 C-RAN Fronthaul design
Whitebox servers allow the virtualization of 5G RAN infrastructure entities by virtualizing some (or all) of the
baseband functions that run on commercial COTS hardware. This approach, called Virtualized RAN (vRAN) is
recognized as a very promising area of innovation in the 5G ecosystem, resulting in cost reductions and
scalability benefits for 5G infrastructure deployments [4]. In our testbed we employ Whitebox servers to
implement all RAN functionalities, i.e., the Remote Radio Head (RRH), and BaseBand Unit (BBU) with open
source software [9], which can be deployed in the form of Virtualized Network Functions (VNFs). Moreover,
our testbed leverages a COTS optical/wireless Fronthaul based on 10 GigE and IEEE 802.11ad, allowing the
eNB functions to be split between the BBU and the RRHs. Multiple RRH units, effectively serving as small-
cells, can be driven by a single BBU which physically resides at the Core site. The RRH units are implemented
with Small Form Factor PCs with commodity IEEE 802.11ad mmWave interfaces that are connected to a USRP
B210 via USB 3.0, which implements the LTE RF functions. As shown in Fig. 2, COTS Ethernet and IEEE
802.11 interfaces are employed to interconnect all entities of the C-RAN architecture, hence avoiding CPRI
interfaces which are only available in carrier-grade, closed ecosystems. Specifically, we employ a 10 GigE
optical fiber link to interconnect the BBU with the RAU, while the RRH-RAU link is implemented with a pair of
commodity IEEE 802.11ad radios. Finally, a FlexRAN Master Controller is employed for the SDN-enabled
centralized control of the RAN. FlexRAN is a flexible and programmable platform which separates the RAN
control and data planes and supports the design of real-time RAN control applications. The Master Controller
controls the underlying RAN infrastructure and orchestrates their operation, facilitating multiple advanced use
cases, such as centralized Resource Block (RB) allocation, load balancing and handover control.
Figure 2: IEEE 802.11ad optical/wireless Fronthaul design
2.2 Core Tier design
Our testbed Core Tier, shown in Fig. 1, employs a virtualized data center based on OpenStack, which hosts
Network Services and Over The Top (OTT) applications. For the mobile Core network, we employ a software-
based EPC from OpenAirInterface project (i.e., openairCN [9]), which includes the implementation of the
Mobility Management Entity (MME), the Home Subscriber Server (HSS), the Service Gateway (S-GW) and the
Packet Gateway (P-GW). The software implementation of both EPC and RAN entities facilitates the concept of
"5G infrastructure as a service" [12], where the RAN and EPC entities can be deployed in the form of VNFs
(i.e., vRRH, vBBU, and vEPC), and hosted by the virtualized data center alongside Network Services and
Applications. The ETSI OSM MANO framework is employed for VNF onboarding and orchestration, allowing
the full automation of operational processes and tasks related to the placement and lifecycle management of all
services. The MANO layer will thus be able to automatically allocate resources to the 5G RAN to accommodate
a new service instantiation, e.g., by deploying new vBBUs, hence alleviating the end user from the burden of
infrastructure management trivialities. Moreover, SDN is a key enabler in our testbed to realize flexible and end-
to-end optimized communication. OpenDaylight controller is employed to centrally control the SDN Switches of
the Data Center (Data plane). The consolidation of the SDN controllers for the wired and wireless domains will
allow us to further pursue Core/RAN orchestration, taking advantage of the global network view from both SDN
controllers. Our current efforts are focused on the implementation of a centralized orchestrator, which works by
consolidating APIs form both OpenDaylight and FlexRAN controllers, implementing SDN East/West (or SDNi)
APIs. This allows the deployment of end-to-end slicing across the heterogeneous infrastructure.
3. Function split types and their effect on mmWave Fronthaul traffic
Our testbed will rely on 60 GHz wireless interfaces compliant with the IEEE 802.11ad specification to prototype
the C-RAN Fronthaul. These devices utilize technology originating from the WiGig MAC and PHY
Specifications, which served as the foundation of the IEEE 802.11ad, bringing to market a new class of very
competitively priced backhauling solutions with multi-gigabit data rates and ultra-low latencies. One of the key
obstacles of C-RAN prototyping is the excessive bandwidth and latency requirements imposed by moving all the
baseband functionality to the cloud. However, migrating some functions to the small-cell site can lead to
significant Fronthaul traffic reduction, allowing cost effective Ethernet and mmWave interfaces to be employed.
On the other hand, as more baseband functionalities are moved to the RRH, the C-RAN advantages are gradually
negated. 3GPP has defined 8 function split types in TR 38.801 [13] ranging from option 8, where all baseband
processing is performed in the cloud, to option 1 where all the baseband processing is performed at the RRH.
OpenAirInterface supports option 8 and option 7 function split types, the latter representing a very efficient
trade-off. In this option, the RRH performs the samples Cyclic Prefix (CP) removal, FFT (i.e., transforming the
samples to the frequency domain) and removing the guard bands, effectively reducing the Fronthaul
requirements by 50%. The core team of the OAI project has estimated that an LTE RRH in a representative
scenario with 20 MHz bandwidth, 4 RX antennas and 64 QAM modulation requires a peak rate of 3.93 Gbps for
an option 8 split type [9]. This is almost halved to 2.15 Gbps for an option 7 split. In both splits, there is a 250 µs
one-way latency constraint [13], which is easily achievable, as IEEE 802.11ad introduced very low PHY
overheads and a low latency CSMA protocol with a 3 μsec SIFS period and PHY headers and preambles with a
duration <2 μsec. As shown in the following table, which details IEEE 802.11ad achievable rates for different
Modulation and Coding Schemes (MCSs) and corresponding SNR values ([14]), the IEEE 802.11ad Fronthaul
can easily serve option 7 split type traffic for MCS8 and higher, which corresponds to an SNR of at least -61
dBm. It must be noted that the peak rate of the option 8 function split type narrowly exceeds the capabilities of
IEEE 802.11ad, which peaks at 4620 Mbps at the highest MCS12, but could potentially be served with a
lightweight compression scheme.
Figure 3: IEEE 802.11ad table of achievable data rates vs. MCS index and SNR
4. CONCLUSIONS
In this work, we have presented the design of an end-to-end 5G platform based on the C-RAN architecture, with
a fully virtualized RAN, an optical/wireless Fronthaul, and a cloud-based backend. Moreover, SDN was
employed for both the wired and the wireless domain, where SDN enables the implementation of centralized
handover and load balancing schemes as well as end-to-end slicing across the heterogeneous network
infrastructure. The main design decisions, i.e., use of open source libraries and COTS components, including
Ethernet and IEEE 802.11ad interfaces to reduce development costs were also discussed. Finally, various
function split types were presented, along with their effect on Fronthaul traffic, and the most appropriate split
types for an IEEE 802.11ad Fronthaul was also discussed.
ACKNOWLEDGEMENTS
This work has been funded by the EC under the auspices of H2020-ICT 5G-PHOS (grant 761989), SPOT5G
(TEC2017-87456-P), H2020-MSCA-RISE CASPER (grant 645393), and H2020-MSCA-ITN-ETN
5GSTEPFWD (grant 722429). Part of this work has been supported by the Generalitat de Catalunya under grant
2017 SGR 891.
REFERENCES
[1] H. Kim and N. Feamster, ‘‘Improving network management with software defined networking’’, IEEE
Commun. Mag., vol. 51, no. 2, pp. 114–119, Feb. 2013.
[2] L. Li, Z. Mao, and J. Rexford, ‘‘Toward software-defined cellular networks’’, in Proc. Eur. Workshop
Softw. Defined Netw., 2012, pp. 7–12.
[3] Chiosi M., et. al., “Network Functions Virtualisation - Introductory white paper,” 2015. [Online].
Available: https://portal.etsi.org/nfv/nfvwhite paper.pdf
[4] “Cloud RAN. Architecture for 5G”. A Telefónica-Ericsson joint White paper. Jul 2016.
[5] T. Huang; F. R. Yu; C. Zhang; J. Liu; J. Zhang; J. Liu, “A Survey on Large-scale Software Defined
Networking (SDN) Testbeds: Approaches and Challenges,” IEEE Commun. Surv. Tuts , vol. 19, no.2,
pp.891-917, 2016.
[6] H. Kim, J. Kim and Y. B. Ko, “Developing a cost-effective OpenFlow testbed for small-scale Software
Defined Networking” in Proc. 16th ICACT, Pyeongchang, 2014, pp. 758-761.
[7] G.Tseliou, F.Adelantado and C.Verikoukis “Scalable RAN Virtualization in Multi-Tenant LTE-A
Heterogeneous Networks”, IEEE Trans. Vehicular Technology, vol. 65, no 8, pp. 6651-6664, Aug. 2016.
[8] S. Costanzo, D. Xenakis, N. Passas and L. Merakos, “OpeNB: A framework for Virtualizing Base Stations
in LTE Networks”, in Proc. IEEE ICC 2014, Sydney, Australia, 10-14 Jun. 2014 pp. 3148–3153.
[9] N. Nikaein, M. K. Marina, S. Manickam, A. Dawson, R. Knopp, and C. Bonnet, “OpenAirInterface: A
flexible platform for 5G research”, SIGCOMM Comput. Commun. Rev, vol. 44, no. 5, pp. 33-38, Oct.
2014.
[10] X. Foukas, N. Nikaein, M. Kassem, M. Marina, K. Kontovasilis, “FlexRAN: A Flexible and Programmable
Platform for Software-Defined Radio Access Networks”, in Proc. 12th CoNEXT, Irvine, CA, USA, 12-15
Dec. 2016.
[11] OpenDaylight: Open Source SDN Platform, Web: https://www.opendaylight.org/.
[12] N. Nikaein, E. Schiller, R. Favraud et al., Towards a Cloud-Native Radio Access Network. Cham: Springer
International Publishing, 2017, pp.171–202.
[13] 3rd Generation Partnership Project; Technical Specification Group Radio Access Network, Study on new
radio access technology; radio access architecture and interfaces, 3GPP TR 38.801 V2.0.0 (2017-03).
[14] C. Y. Chang, R. Schiavi, N. Nikaein, T. Spyropoulos, and C. Bonnet, “Impact of packetization and
functional split on c-ran fronthaul performance,” in Proc. IEEE ICC 2016, Kuala Lumpur, Malaysia, 23-27
May 2016.
[15] J. Kim, J.-K. Kim, “Achievable rate estimation of IEEE 802.11ad visual big-data uplink access in cloud-
enabled surveillance applications”, PLoS ONE, vol. 11, no. 12, Dec. 2016.
