MAB-Based Rate Adaptation with Perfect and Imperfect CSI for IEEE 802.11ax WLANs

Multi-armed Bandit (MAB)-based rate adaptation schemes, Rate Adaptation with General Thompson Sampling (RAGTS), and Rate Adaptation with Sliding-Window Thompson Sampling (RASWTS) are proposed in the literature for IEEE WLAN 802.11ac. The performance of RAGTS and RASWTS is evaluated assuming perfect channel state information (CSI) at the receiver. In MAB-based Rate Adaptation (RA) schemes that rely solely on the MAC layer ACK/ NACK feedback, the receiver's CSI estimation accuracy is crucial in decoding the packet at the receiver. If CSI estimation is imperfect at the receiver, it can lead to packet decoding failures, even when the transmission conditions are favorable. This can cause incorrect NACK feedback, which affects the transmitter's rate adaptation decisions. In this paper, we evaluate and compare the performance of RAGTS and RASWTS under perfect and imperfect CSI conditions for IEEE 802.11ax using standardscompliant MATLAB WLAN Toolbox. Our findings show that CSI estimation errors can lead to incorrect ACK/ NACK feedback, adversely affecting the transmitter's rate adaptation decisions. This results in suboptimal MCS selection and a noticeable degradation in throughput performance.