Antenna selection for massive MIMO Kronecker channel models using non-central principal component analysis

Massive multi-user multiple-input-multiple-output (MU-MIMO) is an emerging fifth-generation (5G) wireless communication technology. However, the deployment of large numbers of antennas and radio frequency (RF) chains needed at the base stations (BS) in massive MIMO brings challenges of high system complexity and hardware energy consumption. By using Kronecker channel modeling, a simple stochastic MIMO model for a narrowband channel is developed that uses the correlation matrices at the mobile station (MS) and BS. In this paper, two semi-heuristic techniques are proposed for practical antenna selection considering Kronecker channel modeling in a MU-MIMO broadcast system using principal component analysis (PCA). Therefore, the number of RF chains needed in massive MU-MIMO is reduced by removing the antennas that contribute least to the system sum capacity. A precoding technique of Zero Forcing is used and each user is equipped with a single antenna. Our simulation results show that the proposed antenna selection technique performs much better than mean channel gain selection.