Hierarchical-Block Sparse Bayesian Learning for Spatial Non-Stationary Massive MIMO Channel Estimation

This letter provides a novel hierarchical-block channel estimation scheme for massive multiple-input multiple-output (MIMO) systems based on Bayesian learning frameworks. Specifically, considering the spatial non-stationarity of massive MIMO channels introduced by a large-scale antenna array, we first exploit a hierarchical-block prior to characterize the delay-domain sparse channel structure across the antenna array and then develop an iterative Bayesian learning scheme to infer the channel vector as well as its hyperparameters associated with the hierarchical-block prior. Meanwhile, the Cramer-Rao bound (CRB) is derived as a reference line. To evaluate our proposed scheme, we modify the standardized 3GPP channel model with birth-death process to capture the spatial non-stationarity of practical massive MIMO channels. Simulation results demonstrate that our proposed scheme can recover channels effectively and attain a significant performance improvement over comparison methods.