Distributed approaches for proportional and max-min fairness in random access ad hoc networks

We consider the question of distributed rate optimization in a wireless random access network, with the goal of achieving global fairness objectives. Since the wireless medium is broadcast in nature, a MAC (Medium Access Control) protocol is essential to reduce collisions and ensure high system throughput. For random access MAC protocols, the feasible rate region is typically a complex, non-convex and non-separable function of the MAC protocol parameters, like the transmission probabilities or back-off window sizes. These factors make the question of distributed rate optimization in a wireless random access network a complex one, and imply that cross-layer coordination is necessary to attain end-to-end fairness. Despite its complexity, significant progress has been made in recent years towards addressing this question for certain fairness measures. In this paper, we summarize our recent results in designing distributed approaches in a random access ad hoc network, with the goal of achieving proportional and max-min fairness at the link layer, and for the end-to-end sessions. We also pose some open questions on this topic that are currently being investigated.