On the performance and fairness of dynamic channel allocation in wireless mesh networks

Channel assignment in multi-channel multi-radio wireless mesh networks is a powerful resource management tool to exploit available multiple channels. Channels can be allocated either statically based on long-term steady state behavior of traffic or dynamically according to actual traffic demands. It is a common belief that dynamic schemes provide better performance; however, these two broad classes of channel allocation schemes have not been compared in detail. In this paper, we quantify the achievable performance gain and fairness improvement through an optimal dynamic channel allocation scheme. We develop optimal algorithms for a dynamic and three static schemes using mixed integer linear programming, and compare them in the context of QoS provisioning, where network performance is measured in terms of acceptance rate of QoS sensitive traffic demands. Our extensive simulations show that static schemes should optimize channel allocation for long-term traffic pattern, and maintain max-min fairness to achieve acceptable performances. Although the dynamic and max-min fair static schemes accomplish the same fairness, the dynamic channel allocation outperforms the static scheme about 10% in most cases. In heavily overloaded regimes, especially when network resources are scarce, both have comparable performances, and the max-min fair scheme is preferred since it incurs less overhead. Copyright c © 2010 John Wiley & Sons, Ltd.