Radio Resource Management for Wireless Mesh Networks Supporting Heterogeneous Traffic

Wireless mesh networking has emerged as a promising technology for future broadband wireless access, providing a viable and economical solution for both peer-to-peer applications and Internet access. The success of wireless mesh networks (WMNs) is highly contingent on effective radio resource management. In conventional wireless networks, system throughput is usually a common performance metric. However, next-generation broadband wireless access networks including WMNs are anticipated to support multimedia traffic (e.g., voice, video, and data traffic). With heterogeneous traffic, quality-of-service (QoS) provisioning and fairness support are also imperative. Recently, wireless mesh networking for suburban/rural residential areas has been attracting a plethora of attentions from industry and academia. With austere suburban and rural networking environments, multi-hop communications with decentralized resource allocation are preferred. In WMNs without powerful centralized control, simple yet effective resource allocation approaches are desired for the sake of system performance melioration. In this dissertation, we conduct a comprehensive research study on the topic of radio resource management for WMNs supporting multimedia traffic. In specific, this dissertation is intended to shed light on how to effectively and efficiently manage a WMN for suburban/rural residential areas, provide users with high-speed wireless access, support the QoS of multimedia applications, and improve spectrum utilization by means of novel radio resource allocation. As such, five important resource allocation problems for WMNs are addressed, and our research accomplishments are briefly outlined as follows: Firstly, we propose a novel node clustering algorithm with effective subcarrier allocation for WMNs. The proposed node clustering algorithm is QoS-aware, and the subcarrier allocation is optimality-driven and can be performed in a decentralized manner. Simulation results show that, compared to a conventional conflict-graph approach, our proposed