Towards Resilient and Reconfigurable Software-defined Wireless Backhaul Networks

Sammanfattning: The increase of mobile devices and services over the last decade has led to unprecedented mobile traffic growth. To cope with the increasing demands, fifth generation (5G) network architectures have been designed to provide the required capacity using a large number of small cells (SCs). However, a dense deployment of SCs requires a robust and scalable backhaul to transport the access traffic towards the Internet. In this thesis, we explore the application of the Software-defined Networking (SDN) paradigm for the management of a wireless backhaul. With SDN, the data and control planes are separated and the network is managed by a centralized entity. To that end, we provide multiple contributions that focus on achieving resilient and reconfigurable wireless backhaul networks. Firstly, we propose an SDN-based architecture to manage the wireless backhaul. Our architecture is integrated in practical testbed environments, where we use an SDN controller to configure the forwarding plane and wireless backhaul links. Secondly, we evaluate SDN-based resiliency in the wireless backhaul. We achieve that by implementing fast-failover resiliency with OpenFlow group tables and by using the bidirectional-forwarding detection protocol (BFD) to monitor the state of the backhaul links. Finally, we develop algorithms that calculate the necessary reconfiguration operations to transition between different wireless backhaul topologies, while minimizing the impact on existing user traffic. We consider that the backhaul nodes can be powered on/off and are equipped with steerable antennas that can be aligned to form links with different neighbors. Our optimization problems are modeled as mixed integer linear programs (MILP) that are optimally solved using exact mathematical programming methods. In addition, we develop greedy-based heuristic algorithms that solve the same problems and obtain good quality solutions in short time.