Rate Adaptation Algorithm with LSTM in IEEE 802.11ac

IEEE 802.11 WLANs are widely used in various industries for broadband data transmission, such as HD photos or videos captured by cameras, mobile robots, or unmanned aerial vehicles (UAV) in power substations and transmission lines. The rate adaptation algorithm (RA algorithm) is crucial in determining the transmission performance of 802.11 WLANs. An effective RA algorithm can achieve higher throughput and lower communication delay for the system. However, due to the increasing complexity of the MCS table in the 802.11 standard, it is challenging for the RA algorithm to adapt to changing channel quality caused by mobility of nodes or wireless interference in a timely manner. In this paper, we propose a new RA algorithm called LSTM-MinstrelHt, based on the MinstrelHt algorithm. This algorithm uses Long Short Term Memory network (LSTM) to predict the SNR value and selects a reasonable MCS index based on the predicted SNR to accelerate the convergence of the RA algorithm. By simulating on network simulator 3 (ns-3), our results show that the LSTM-MinstrelHt algorithm can improve the average throughput by 11% and reduce the average delay by 13% compared to the original algorithm.