DDPG-based Multi-AP Cooperative Access Control in Dense Wi-Fi Networks

Multi-Access Point (AP) cooperation is one of the potential core technologies in the future Wi-Fi 8 standard. By sharing information among multiple APs, it can improve spectrum efficiency and throughput in dense scenarios. However, most of existing researches on Wi-Fi access control mechanism optimization focuses on adjusting the contention window (CW) and carrier sense threshold (CST) values within a single Basic Service Set (BSS), which do not well support the multi-AP cooperation for densely deployed multi-BSS scenarios. The overlapping BSS preamble-detection (OBSS PD) defined in the 802.11ax standard is the basis of spatial reuse (SR) in OBSS scenarios. It only provides the adjustment range and constraints for OBSS PD, but does not specify how to adjust OBSS PD. Moreover, there is a certain coupling relationship between CW and OBSS PD. In order to efficiently carry out the multi-AP cooperation in dense Wi-Fi scenarios, in this paper, we propose an AI-enabled optimization algorithm for multi-AP access control based on a deep reinforcement learning method-the Deep Deterministic Policy Gradient (DDPG) method, which jointly adjusts the CW and OBSS PD parameters through multi-AP cooperation. The goal is to improve the aggregate throughputs in OBSS scenarios. The simulation results indicate that the proposed algorithm can improve throughput performance by approximately 7.91% to 56.85% compared to several baseline schemes.