Performance of Cell-Free Massive MIMO with Joint User Clustering and Access Point Selection

We consider an uplink cell-free massive multiple-input multiple-output (MIMO) system, in which the access points are connected to the central processing unit (CPU) through a fronthaul network. This system has the advantages of wide coverage and flexible deployment. However, the performance of this system depends on a capacity-limited fronthaul, and when the fronthaul is saturated, the quality of service will be reduced. To address this issue, we propose a joint user clustering and AP selection scheme, which can reduce the pressure on the fronthaul link while taking into account the system performance and computational complexity. We first derive a closed-form expression for the uplink spectral efficiency over Rician fading channels. Based on the derived expression, we formulate the problem of maximizing the minimum uplink spectral efficiency across all users by jointly optimizing the large-scale fading decoding (LSFD) coefficient and power control coefficient. Then, combined with the optimization results and channel estimation error, a suboptimal access point selection scheme is proposed. In addition, we propose a user clustering scheme to further reduce the complexity of the AP selection scheme. The simulation results show that the joint user clustering and access point selection scheme can reduce the system fronthaul link pressure, while the performance degrades only slightly.