Joint Access Points-User Association and Caching Placement Strategy for Cell-Free Massive MIMO Systems Based on Soft Actor-Critic Algorithm
Rui Wang, Min Xuan Shen, He Yun, Xiangyan Liu · IEEE Communications Letters · 2023
The joint optimization of APs-user association (AUA) and caching placement (CP) presents a promising method to mitigate the fronthaul overhead and reduce user-perceived latency in cell-free (CF) massive MIMO systems. However, previous studies have not considered the impact of deploying multiple access points (APs) and the user-centric nature of CF massive MIMO systems on joint schemes, while also lacking flexibility in managing caching and association resources. In this letter, we initially develop a latency model based on the collaborative caching and user-centric characteristic. We then formulate a long-term joint optimization problem for AUA and CP to minimize content delivery latency. A two-timescale soft actor-critic (SAC) deep reinforcement learning based algorithm is proposed to solve this NP-hard problem. To enhance the convergence of the SAC algorithm, we devise user clustering and cache partitioning schemes. Numerical results demonstrate that our proposed scheme exhibits reduced latency compared to the benchmark schemes, while efficiently utilizing fronthaul and caching resources.