Hybrid Precoding Based on MMSE-EDS for Multi-user Multi-stream Massive MIMO Systems

Abstract

Hybrid analog/digital precoding architecture is a combination of analog and digital processing that reduces the number of RF chains while enabling beamforming and spatial multiplexing gain. Existing mean squared errors (MSE) hybrid precoding algorithm can minimize the sum of MSE, however, do not guarantee that the MSE of each data stream minimized. In this paper, we propose a hybrid precoding method based on the minimum mean square error of each data stream (MMSE-EDS) for multi-user multi-stream Massive multi-input multi-output (MIMO) system. Through calculating the optimal baseband precoding and analog precoding pairs for each data stream, the proposed hybrid precoding ensures minimization of the MSE of each data stream, thus minimizing the sum of the mean square error (MSE) of all data streams. The proposed precoder/combiner can be easily obtained by Lagrangian algorithm. The performance advantages of our proposed hybrid precoding in spectral efficiency (SE) and bit error rate (BER) have been simulated.