Quantized Variational Bayesian Joint Channel Estimation and Data Detection for Uplink Massive MIMO Systems with Low resolution ADCS

Abstract

In this paper, we consider the joint channel estimation and data detection in an uplink massive multiple input multiple output (MIMO) receiver with low resolution analog to digital converters (ADCs). The nonlinearities introduced by the ADCs make the existing linear multiuser detection (MUD) approaches suboptimal, and motivates a fresh look at the problem. Also, channel state information is necessary to obtain the channel quality metrics that are used for link adaptation by the base station (BS). We model the MIMO receiver system as a directed probabilistic graphical model, and propose a variational Bayesian procedure to estimate the channel and the posterior beliefs of the transmitted symbols. We evaluate the symbol error probability (SEP) and the normalized mean squared error (NMSE) of the channel estimates of the proposed algorithm using Monte Carlo simulations, and benchmark it against an unquantized variational Bayesian algorithm with perfect and imperfect channel state information at the receiver (CSIR).