Hybrid Precoding Design in Multi-User mmWave Massive MIMO Systems for BER Minimization

The hybrid precoding technology can be utilized to reduce the bit error rate (BER) of millimeter wave (mmWave) massive multiple-input multiple-output (MIMO) systems. Nevertheless, most prior hybrid precoding schemes aiming to optimize the BER focus on the conventional minimum mean square error (MMSE) criterion, which can achieve limited BER performance. In this letter, by analyzing the average BER expression, we find that the average BER of the multi-user (MU)-MIMO system will achieve a lower bound when MSEs among streams of different users satisfy certain constraints. Based on this point, we propose a constrained MMSE (CMMSE) criterion to facilitate designing the hybrid precoding for minimum average BER. Moreover, a two-stage design is proposed to tackle the CMMSE problem, where an effective singular vectors approximation (SVA)-based approach is proposed to construct the analog precoder/combiners in the first stage and the MSE constraints are fully considered for the digital precoder design in the second stage. The superior BER performance of our proposal over alternatives is demonstrated by simulations.