Fair scheduling with bottleneck consideration in wireless ad-hoc networks

Most research work in the area of wireless ad-hoc networks attempts to balance the trade-off between fairness and channel utilization. In this paper, we first propose a topology-independent methodology to predict maximum achievable channel utilization under fairness constraint by two performance bounds. Based on the notion of bottlenecks introduced in prediction, we design a centralized and improved fair scheduling algorithm for wireless ad-hoc networks. We capture traffic load characteristics by using a proposed parameter that represents the "contending power" of nodes in the weighted flow contention graph. Finally, we demonstrate the effectiveness of our proposed algorithm through both provable analysis and simulations, and discuss natural derivations of a fully distributed algorithm using our bottleneck-based analytic model