Abstract: Cell-free Massive MIMO achieves uniform network coverage and eliminates cell-edge effect by dispersive deployment of large number of access points (APs). In cell-free Massive MIMO system, APs cooperate to provide services for users by interacting with the central processing unit through the fronthaul link. To avoid the overloading of the fronthaul link caused by all APs simultaneously serving all UEs, the AP selection strategy in CF M-MIMO is studied. For this purpose, the average spectral efficiency per link is defined as the spectral efficiency obtained per AP-UE link, which is used to measure the effectiveness of the access point selection algorithm. Considering user quality of service (QoS) requirements, an access point selection scheme based on the Deep Deterministic Policy Gradient (DDPG) algorithm is proposed. The simulation results show that compared with the existing access point selection scheme, the proposed DDPG-based access point selection scheme can effectively improve the average link spectral efficiency of the CF M-MIMO system and meet the user's QoS requirements.

Key words: Massive MIMO, cell-free network architecture, access point selection, DDPG