Effects of Retransmission Limit on the Performance of Centralized Random Backoff

Abstract

Previous studies showed that the packet collision probability in the 802.11 Distributed Coordination Function (DCF) can approach zero by adopting Centralized Random Backoff (CRB), which improves the total throughput performance. Moreover, it was reported that wireless networks using CRB maintain a high level of compatibility to legacy wireless local area networks (WLANs) when they coexist. However, little attention has been paid to relations between the performance of CRB and the retransmission limit, although the number of retransmission limit is an essential parameter that must be carefully chosen when implementing CRB for efficient WLANs. This article presents more realistic Markov chain models, and studies the impact of retransmission limit on the performance of CRB. These relations are also investigated by Monte Carlo simulation in more practical scenarios. Results show the trade-off that the total throughput increases with the retransmission limit when the number of nodes is larger than 15; whereas, the channel access opportunities in the mixed network scenario are more unevenly distributed between nodes as the retransmission limit increases. It is also shown in the simulation results that the throughput degradation caused by the hidden node problem is alleviated as the retransmission limit increases.