Optimising channel assignment to prevent flow starvation and improve fairness for planning single radio WMNs in built environments

Wireless mesh networks (WMNs) have many attractive characteristics, such as auto-configuration, self-management, and self-healing. With newer and farther reaching applications being developed in built environments, such as smart grids and intelligent transportation systems, users expect high quality of service and thus fairness is an important issue to be addressed. Channel assignment (CA) is the mechanism for allocating radio resources to the nodes and therefore plays a key role in managing fairness in WMNs. Fairness in WMNs depends on how wireless resources are allocated among the nodes. We examine interference models used in existing CA algorithms and find that CA algorithms using these models yield poor fairness because they only reflect local interference between a link and its interfering links. However, flow starvation is due to network wide interference (i.e. global) involving border links and middle links. We propose a novel anti-starvation channel assignment algorithm (ASCA) for planning single radio WMNs. Such ASCA algorithm leverages a new interference model that takes into account both local and global interference. Simulation results show the ASCA algorithm effectively alleviates flow starvation and improves fairness up to 62% compared with the best result from clique-based CA benchmarks. To the best of our knowledge, the proposed ASCA is the first one to optimise CA algorithms with consideration of both local and global interference.