Department of Electronic and Electrical Engineering, The University of Sheffield, Sheffield, South Yorkshire S10 2TN, UNITED KINGDOM.
In the past few years, severe challenges have arisen for network operators, as explosive growth and service differentiation in data demands require an increasing number of network capacity as well as dynamic traffic management. To adapt to the network densification, wireless backhaul solution is attracting more and more attentions due to its flexible deployment. Meanwhile, the software-defined network (SDN) proposes an promising architecture that can achieve dynamic control and management for various functionalities. In this case, by applying the SDN architecture to wireless backhaul networks, the traffic scheduling functionality may satisfy the ever-increasing and differentiated traffic demands.
To tackle the traffic demand challenges, traffic scheduling for software-defined backhaul networks (SDBN) is investigated from three aspects in this thesis. In the first aspect, various virtual networks based on service types are embedded to the same wireless backhaul infrastructure. An algorithm, named VNE-SDBN, is proposed to solve the virtual network embedding (VNE) problem to improve the performance of the revenue of infrastructure providers and virtual network request acceptance ratio by exploiting the unique characteristics of SDBNs. In the second aspect, incoming traffic is scheduled online by joint routing and resource allocation approach in backhaul networks operated in low-frequency microwave (LFM) and those operated in millimetre wave (mmW). A digraph-based greedy algorithm (DBGA) is proposed considering the relationship between the degrees of vertices in the constructed interference digraph and system throughput with low complexity. In the third aspect, quality-of-service is provided in terms of delay and throughput with two proposed algorithms for backhaul networks with insufficient spectral resources. At last, as a trial research on E-band, a conceptual adaptive modulation system with channel estimation based on rain rate for E-band SDBN is proposed to exploit the rain attenuation feature of E-band.
The results of the research works are mainly achieved through heuristic algorithms. Genetic algorithm, which is a meta-heuristic algorithm, is employed to obtain near-optimal solutions to the proposed NP-hard problems. Low complexity greedy algorithms are developed based on the specific problem analysis. Finally, the evaluation of proposed systems and algorithms are performed through numerical simulations. Simulations for backhaul networks with respect to VNE, routing and resource allocation are developed.