Turbo-Like Joint Data-and-Channel Estimation in Quantized Massive MIMO Systems

We consider joint channel-and-data estimation for quantized massive MIMO systems. The estimation for both parts follows a turbo-like fashion, where the estimation error of one step is treated as additive Gaussian noise for the other. An approximate belief propagation algorithm is employed to obtain an approximate minimum mean square error estimate of both the data and channel. The performance of our scheme is compared to a Bayes optimal joint channel-and-data estimation approach by Wen et al. (2015). We observe that 10 turbo iterations are enough to achieve similar performance with lower complexity.