Wireless Back-haul: a software defined network enabled wireless Back-haul network architecture for future 5G networks

Christian Niephaus, Senka Hadzic, Osianoh Glenn Aliu, Gheorghita Ghinea, Mathias Kretschmer
2015 IET Networks  
Recently both academic and industry worlds have started to define the successor of Long Term Evolution (LTE), so-called 5G networks, which will most likely appear by the end of the decade. It is widely accepted that those 5G networks will have to deal with significantly more challenging requirements in terms of provided bandwidth, latency and supported services. This will lead to not only modifications in the access segment and parts of core networks, but will trigger changes throughout the
more » ... e network, including the Back-haul segment. In this work we present our vision of a 5G Back-haul network and identify the associated challenges. We then describe our Wireless Back-haul (WiBACK) architecture, which implements Software Defined Network (SDN) concepts and further extends them into the wireless domain. Finally we present a brief overview of our evaluation results. Introduction It is widely accepted that future 5G networks, which are expected by the end of the decade, have to cope with a huge variety of existing and novel services and applications, such as cloud-based applications, ultra-HD television and online conferencing, Machine-Type-Communication (MTC) as well as augmented reality [1]. First and foremost, this leads to a tremendous increase in the required capacity [2] for future networks. Besides that, 5G networks also need to provide high reliability and very low latency in order to support those novel applications and allow for a good Quality-of-Experience (QoE) perceived by the user. Precisely, 5G networks are supposed to provide 10x higher throughput per User Terminal (UT) and 1000x more traffic in the Back-haul network connecting the Base Stations (BSs) [3], while reducing the service level latency to 5ms and maintaining a reliability of 99.999% [4] . A lot of effort is made to investigate novel technologies and mechanisms for the access networks, such as mmWave, ultra dense deployments or massive multiple-input and multiple-output (MIMO), in order to cope with the aforementioned requirements, yet the implications on the Back-haul segment of the network are often silently ignored. However, the increasing number of cells, each supporting a significantly higher throughput, needs to be adequately connected to transport the traffic to the core segment and eventually the Internet. In existing 4G networks this is usually not an issue, since Evolved Node Bs (eNodeBs) are either connected with a high speed wired connection, e.g. optical fiber, or with a high capacity micro-wave link so that the Back-haul network typically does not present a bottleneck. With the advent of network densification and small cells, which are deployed at non-conventional locations, more complex Back-haul network structures relying often on heterogeneous wireless technologies will appear. For example, cells might be located on traffic lights, distributed in-and outdoor on large campuses, etc., where a wired infrastructure and high capacity line of sight (LOS) microwave links are either impractical or too costly to deploy everywhere. Instead, more cost-effective wireless connections might be used, most likely even in a multihop fashion. Hence, an over-provisioned Backhaul network, as it is the usual case in 4G networks, becomes rather the exception and capacity constraints will occur. In order to moderate those, intelligent and self-organizing Back-haul techniques are required, which utilize the available spectrum most effectively by avoiding interferences with each other and the Radio Access Network (RAN). Moreover, in order to provide sufficient bandwidth virtually everywhere including rural and remote areas, satellite overlay networks will be used in future 5G networks. Satellite networks are expected to evolve tremendously in recent years into a terrabit per second communication system, yielding to a significant decrease in costs per bit [5] . They are primed for broadcast and multicast type of services covering large areas and are therefore particularly suited to provide additional capacity wherever it is needed. Furthermore, the Back-haul network might be equipped with additional storage capabilities, i.e. caches, or even computational units. This is required to bring content and services closer to the user in order to achieve the demanded low latency. Thus, the Back-haul network needs to become highly flexible in terms of traffic controlling, capacity management as well as radio management and therefore heavily rely on novel SDN, Network Function Visualization (NFV) and cloud concepts [15] .
doi:10.1049/iet-net.2015.0009 fatcat:bl3b6mk4vna3plembq42qgr63e