Study on cluster formation for distributed MU-MIMO

Deploying a large number of spatially distributed antennas (DAs) is a promising solution to improve the link capacity while keeping the coverage of 5G systems which utilize a high frequency band. And clustering is a promising method to mitigate the extremely high computational complexity due to a large-scale multi-user multiple-input multiple-output (MU-MIMO). User-clustering and antenna-clustering are two clustering approaches. In this paper, we utilize K-means algorithm in clustering and propose two cluster member assignment (CMA) methods to assign corresponding DAs or users into user-clusters or antenna-clusters. Then, we evaluate the achievable downlink and uplink capacities through computer simulation to find the optimum combination of clustering approach and CMA.