Traffic scheduling for QoS provisioning in broadband wireless access networks

The unprecedented growth of Broadband Wireless Access (BWA) networks in recent years has boosted the use of both established and new multimedia applications. In order to satisfy the diverse Quality of Service (QoS) requirements entailed by these applications, an efficient traffic scheduler is deemed crucial. At present, a substantial amount of work on scheduling has been done, and those algorithms that are equipped with channel-aware capability appear to be very promising for wireless networks. These schedulers take advantage of the temporal variations of channels and optimize wireless resources accordingly. Nevertheless, there is still lack of clarity as to which channel-aware scheduling algorithm is the most suitable to support QoS in a heterogeneous traffic environment. Furthermore, most of the prior research work focuses on the downlink direction. The increasing popularity of interactive multimedia applications calls for an urgent need to study the uplink case as well.;This thesis presents a methodology for extending the existing wireless scheduling algorithms to cope with the heterogeneous traffic environment, particularly in the uplink direction. More specifically, the Weighted Round Robin (WRR) scheduling and the combination of the Modified Largest Weighted Delay First and Proportional Fair (M-LWDF-PF) scheduling are considered, and their effectiveness in providing QoS differentiation for multiple traffic types is investigated. Derived from the M-LWDF-PF scheduling, a dynamic scheme named beta-M-LWDF-PF is then proposed, which adapts the service differentiation based on the online network load measurements. The proposed scheme utilizes the reported queue state information of uplink flows to dynamically adjust the service differentiation between real-time and non-real-time services, offering a more flexible QoS differentiation mechanism that works well with various network load conditions. From the simulation results, the proposed beta-M-LWDF-PF scheme shows improvement in the delay performance for real-time users, while maintaining reasonable throughput for non-real-time and best effort traffic. In comparison with the original M-LWDF-PF scheme, the proposed scheduling scheme can more effectively accommodate the demands of real-time users in the mixed service scenarios.