Opportunistic Random Access in CSMA/CA-Based Wireless Networks

Abstract

IEEE 802.11 MAC based CSMA/CA (Carrier Sense Multiple Access with Collision Avoidance) is the most widely used protocol in wireless networks. In CSMA/CA, every node listens to the shared wireless medium and transmits only when the channel is sensed idle to avoid possible collisions. CSMA/CA allows each user of equal probability in access- ing wireless channel, which incurs equal throughput in long term regardless of the channel conditions. To exploit user diversity that refers to the di erence of channel condition among users, we proposed three opportunistic random access mechanisms: overlapped contention, segmented contention and normal distribution based contention, to favor the user of best channel condition in channel access. In the overlapped contention, the contention windows of all users share the same ground of zero, but have di erent upper bounds upon channel condition. In the segmented contention, the contention window upper bound of a better channel condition is smaller than the lower bound of a worse channel condition; namely, their contention windows are segmented without any overlapping. In the normal distribu- tion based contention, the back-o interval is determined using normal distribution with the expectation of proper mean and standard deviation value within the contention window. These algorithms are also enhanced to provide temporal fairness and avoid starving the users with poor channel conditions. The proposed mechanisms are implemented and eval- uated in the NS3 simulator. Extensive experiments show that the proposed opportunistic random access schemes can signi cantly improve the network performance in throughput, delay, and jitter over the current CSMA/CA in IEEE 802.11 networks. In particular, the overlapped contention scheme can o er 73.3% and 37.5% throughput improvements in the infrastructure-based and ad-hoc networks, respectively.