Cooperative Pseudo-Bayesian Backoff Algorithms for Unsaturated CSMA Systems with Multi-Packet Reception

This paper proposes efficient backoff algorithms for uplink multi-packet reception (MPR) capable IEEE 802.11 systems in order to maximize the system throughput. According to the proposed algorithms, each station (STN) estimates, in a Bayesian manner under an unsaturated channel traffic condition, the number of backlogged STNs sharing the multiple access channel to obtain an optimal (re)transmission probability. Additionally, an access point and associated STNs cooperate by exchanging information piggybacked in transmitted data packets and the corresponding acknowledgment packets. The mean and variance of the queuing delays of the proposed algorithms are extensively evaluated via simulations under various environments such as time-varying populations and various asymmetric traffic conditions and compared to those of the conventional binary exponential backoff (BEB) algorithm. Furthermore, the queuing performance of the proposed algorithms is compared to the queuing delay lower bound obtained from a system that has perfect knowledge of the backlog size. Numerical results demonstrate the robustness of the proposed algorithms in various environments, and that they outperform the BEB algorithm.