On reducing delay and temporal starvation of queue-length-based CSMA algorithms

Recently, a group of queue-length-based CSMA algorithms have been proposed to achieve throughput optimality in wireless networks with single-hop transmissions. These algorithms suffer from two problems: (1) large delays, and (2) temporal starvation phenomenon, where communication links are inactive for a prolonged period of time before getting service. To mitigate these two problems, in this paper, we propose a novel v(t)-regulated CSMA algorithm which can be implemented in a distributed manner using the RTS/CTS mechanism. Link scheduling is performed such that links with longer queues are favored so as to reduce average delay. The v(t)-regulated CSMA algorithm also ensures a more frequent switch between schedules such that the effect of temporal starvation is reduced. We prove that the proposed algorithm is throughput optimal, and show through numerical evaluations that the algorithm indeed mitigates the temporal starvation problem and achieves far better delay performance than other throughput-optimal CSMA algorithms.