Binary exponential backoff based congestion control mechanism in multihop wireless networks

In multihop wireless networks, congestion mainly due to the medium contention in MAC layer, which is quite different from traditional wired networks. This paper presents a solution based on the binary exponential backoff algorithm, namely, MAC layer congestion control (MLCC) mechanism. As it is known, a node needs to generate a random backoff period each time when it failed to transmit a frame. Thus the more frequently one node generates the backoff period, the more seriously the network congests. In MLCC, we use this information as a metric for network congestion, thus a node can adaptively adjust its packet loss probability to alleviating MAC collisions and network congestion. Finally, this paper proposes a network throughput function and uses it to prove the validity of the novel congestion metric. Simulation results show that the MLCC algorithm effectively alleviate the network congestion, and significantly increase the network throughput.