A Novel Collision Avoidance Algorithm for IEEE 802.11 Wireless LANs

As the contention resolution scheme in IEEE 802.11, binary exponential back off (BEB) has long been criticized for its high collision probability in saturation situation. In this paper, a novel contention resolution algorithm is designed to overcome its disadvantage. The cause of BEB's poor performance is identified as the overlapped back off intervals of stations and the restoration of contention window to the smallest value upon a successful transmission. Therefore two new concepts, namely back off interval isolation (BII) and improved slow decrease (ISD), are proposed in our enhanced collision avoidance (ECA) algorithm. BII introduces an extra back off interval selection process and stations at different stages have separated back off intervals. With ISD, contention window can only decrease after a certain number of successive successful transmissions. Two dimensional Markov chain is utilized to model the contention resolution process and the theoretical performance of ECA is also deduced. Simulation results show that our algorithm outperforms BEB in both throughput and delay by lowering the collision probability.