Self-Learning Rate Adaptation in Multi-Rate 802.11 Networks

Given the multi-rate option in most 802.11 standards, rate adaptation based on dynamic channel condition is crucial to the system performance. This paper presents an efficient self-learning rate adaptation mechanism at MAC layer, called SLRA. SLRA takes both the temporal or spatial correlation and change into account. The rate selection has two steps: one is to evaluate the performance of current rate and decide whether to adjust the rate; the second is to calibrate the decision according to the historical experiments of rate adjustment at similar channel status. We evaluate SLRA in 802.11 network and find that it can choose the right rate at most time and performs better than other sender-side adaptation schemes.