Resource Allocation for User-Centric Cell-Free Massive MIMO Downlink

Given the strong macro-diversity gain, cell-free massive multiple-input multiple-output (MIMO) with many access points (APs) can provide a more uniform service level for randomly distributed users or user sensors (USs). There has, however, been little implemented in practical applications, due to the challenges of substantial backhaul overhead and high signal processing demands in cell-free massive MIMO systems. In this paper, we propose a resource allocation scheme that incorporates the effective algorithms for AP selection, pilot assignment, and power allocation, to improve the performance of user-centric cell-free massive MIMO systems. In particular, a grouping-based AP selection algorithm is investigated to select APs with significant gain contributions to USs. Based on this, we propose an improved K-means clustering (I-KMC)-based pilot assignment algorithm to reduce pilot contamination caused by pilot reuse. To improve the reliability of data transmission, we present two power allocation algorithms to maximize the achievable sum-rate, namely sequential convex approximation in combination with geometric programming (SCA-GP) approach, and nonmonotone accelerated proximal gradient (NAPG) approach with line search. Results demonstrate that the proposed scheme achieves similar sum-rate performance to the scenario where all USs are served by all APs, but with significantly lower power consumption.