Death by Starvation?: backhaul and 5G

CTN Issue: September 2015

Alan Gatherer, Editor in Chief, ComSoc Technology News

Editors note: Our continued quest to find 5G's weakest link has brought us to backhaul. This month Henrik Lundqvist of Huawei, Sweden has kindly agreed to lead us through the pros and cons of backhaul for 5G. This is a warm-up for the upcoming Special Edition of IEEE Wireless Communications Magazine, for which he is also an editor.

Henrik Lundqvist, Senior Researcher, Huawei Sweden.

The increasing data volume in mobile networks is a challenge not only for the radio access interface but also for the backhaul. Looking ahead to 5G there are different approaches being proposed to increase the network capacity, including massive MIMO, usage of larger spectrum bands and densification of the network. Regardless which of these solutions are used, the backhaul will need to transport more data; but from an economical and technical perspective, densification is more challenging as it requires the backhaul to reach more access node sites. For the access interface, the shorter distance between the mobile device and the access node increases both the signal-to-noise ratio and the available spectral resources. This has been the main contribution to increased network capacity over the years as observed in Cooper’s law (http://www.arraycomm.com/technology/coopers-law/). However, all the access nodes in this dense network need to be connected to a core network or the Internet, which means that some of the technical challenge is moving from the access link to the backhaul. This is something we are already used to from WLANs, where the bottleneck is often the connection to the Internet rather than the radio interface. Mathematical modeling has also shown that the backhaul delay and cost put limits on how densely it makes sense to deploy small cells in a heterogeneous network [1]. Therefore, the backhaul limitations need to be considered in the system design for 5G.

The requirements on the backhaul depend on the network implementation. In particular there is a strong interest in implementing cloud-RAN (C-RAN), with centralized baseband processing [2]. C-RAN enables more advanced joint processing to handle interference, and it also allows pooling of computing resources between multiple radio heads. Transporting soft bits (i.e. samples quantized in to multiple levels) to the baseband processing units is referred to as fronthaul rather than backhaul. The requirements of fronthaul are more difficult to meet due both to the increased data rates that are required to transport soft bits rather than detected data and due to the low latency required for the processing. Already for LTE the required bitrates are on the order of hundreds of Gb/s and the one way latency requirement considered by the Next Generation Mobile Networks (NGMN) Alliance is 250 microseconds [3]. Since 5G targets both shorter latency and higher throughput the bar will be raised even further. The Common Public Radio Interface (CPRI) has been defined to support interoperability between different vendors for fronthaul [4]. Generally, different transport solutions over optical fiber seem most promising when it comes to such large capacity requirements [5]. However, in a dense deployment it is difficult to find enough sites for access nodes that can be connected by fiber at a reasonable cost. Therefore, it may be necessary to rely on wireless backhaul or fronthaul for many of the sites [6]. Efficient integration of wireless, fiber and possibly other transport solutions are important for a cost-efficient deployment.

To handle the constraints from backhaul/fronthaul links with varying characteristics, some flexibility in the implementation is needed. The requirements on the fronthaul can be changed depending on how the digital processing is split between data centers and remote radio units. Depending on the available fronthaul capacity and latency it may therefore be necessary to make the function split so that more of the processing is done in the RRU to relax the fronthaul requirements. In 5G this might even be done dynamically and adapted to specific application requirements, relying on software-defined networking and network function virtualization.

With a wireless backhaul or fronthaul connecting many small base stations, many solutions that are considered for the access interface are also of interest for the backhaul. For example, higher frequency bands that have so far been used for point-to-point backhaul radio links are being considered to be used for the access link. To allow more flexible topology and simpler installation, beamforming based on large scale MIMO is also being considered for the backhaul links. Since the RRUs are in general stationary, the challenges of tracking a mobile user and frequently estimating the channel are relaxed compared to the access link, where these are major problems for millimeter wave and massive MIMO. Because the technology between access links and backhaul links will be similar it may be possible for much of the interface to be the same. It is already possible in LTE to use relays;  in future systems the efficiency could be improved for example if the relay uses full duplex transmission within the same frequency band. A more limited approach to integrate backhaul and access is to take the backhaul limitations into consideration in the radio resource management algorithms, even if the backhaul uses separate resources.

To summarize, a denser deployment of access nodes moves some of the challenges from the access interface to the backhaul. Therefore, the backhaul needs to be taken into closer consideration than had been the case in previous cellular systems.  In October,  a special issue of  IEEE Wireless Communications Magazine will appear where many aspects of backhaul and fronthaul for 5G networks are treated. This demonstrates that this is currently becoming an important question in both industry and the research community. For those interested in the topic, the special issue will give both a good overview of the problems and a number of exciting articles proposing solutions for the backhaul and fronthaul for 5G. Despite being challenging the backhaul is not likely to be a show stopper for 5G, it just has to be given sufficient attention.

References:

[1] D. Chen, T. Quek, M. Kountouris, “Backhauling in Heterogeneous Cellular Networks: Modeling and Tradeoffs”, IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, VOL. 14, NO. 6, JUNE 2015

[2] NGMN Alliance, FURTHER STUDY ON CRITICAL C-RAN TECHNOLOGIES, March 2015.
https://www.ngmn.org/uploads/media/NGMN_RANEV_D2_Further_Study_on_Critical_C-RAN_Technologes_v1.0.pdf

[3] NGMN Alliance, FRONTHAUL REQUIREMENTS FOR C-RAN, March 2015. https://www.ngmn.org/uploads/media/NGMN_RANEV_D1_C-RAN_Fronthaul_Requirements_v1.0.pdf

[4] CPRI Cooperation. CPRI specification v6.1: Interface specification, July 2014.
http://www.cpri.info/downloads/CPRI_v_6_1_2014-07-01.pdf

[5] NGMN Alliance, BACKHAUL AND FRONTHAUL EVOLUTION, March 2015.
https://www.ngmn.org/uploads/media/NGMN_RANEV_D4_BH_FH_Evolution_V1.01.pdf

[6] Skyfiber Inc., “Breaking the Backhaul Bottleneck,” White Paper, June 2013.
http://www.skyfiber.com/assets/docs/pdf/technologyandresouces/WP_SkyFiber_Mobile_Backhaul.pdf

Editor-in-Chief: Alan Gatherer (alan.gatherer@huawei.com)

Comments welcome!

Leave a comment

Statements and opinions given in a work published by the IEEE or the IEEE Communications Society are the expressions of the author(s). Responsibility for the content of published articles rests upon the authors(s), not IEEE nor the IEEE Communications Society.