Density-Based Clustering Two-Stage Precoding Scheme in Massive MIMO System

We propose a density-based clustering two-stage precoding scheme to solve the problems of non-channel reciprocity and high complexity of precoding of downlink channels in an FDD Massive MIMO system. After dividing the users with equivalent of channel autocorrelation matrix into one group, we perform two-stage precoding for each group to reduce system overhead. The numerical results demonstrate that compared with the traditional linear precoding algorithm, the proposed scheme greatly reduces the complexity of the system, achieves higher system capacity and lower BER.