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.