AoA Estimation-Aided Bayesian Receiver Design via Bilinear Inference for mmWave Massive MIMO

This paper proposes a novel angle-of-arrival (AoA) estimation-aided Bayesian joint channel and data estimation (JCDE) algorithm for uplink signal detection in millimeter-wave (mmWave) massive multi-user MIMO (MU-MIMO) systems with short non-orthogonal pilots. In the proposed method, the prior distribution of mmWave channels in the angular domain after digital beamforming is approximated by a Bernoulli-Gaussian (BG) distribution, and the mismatch with the actual distribution is corrected based on AoA estimation by low-complexity angle rotation (AR) method. The resultant beam-domain JCDE algorithm has an inherent mechanism to update path gains of the channels estimated based on AoA for every iteration. This receiver design allows us to take the advantage of both stochastic (i.e., Bayesian) and deterministic (i.e., AR) approaches. Efficacy of the proposed method over the state-of-the-art is confirmed via computer simulations in terms of bit error rate (BER) performance compared to the state-of-the-art alternatives.