Modeling CSMA/CA network under asymmetric conditions

Abstract

Prior work reveals that unfairness in CSMA networks is mainly caused by three asymmetries: (i) information asymmetry; (ii) channel asymmetry; and (iii) rate asymmetry. In this paper, we develop a simple yet accurate analytical model that jointly characterizes all three asymmetric conditions. Our model allows us to estimate the throughput of two contending flows, given the contention window (CW) values. The model can be applied to remedy unfair issues in these topologies, by enabling CW tuning mechanisms to quantify the resulting throughput. The model is flexible enough to support three different fairness objectives. Further, we illustrate how to apply our CW tuning mechanism to an extensive set of asymmetric topologies. Using simulations and testbed measurements, we show that our CW tuning mechanism is effective in mitigating unfairness in the presence of the three fundamental asymmetric conditions.