Data-Aided Dual-Space Channel Estimation Resilient to Pilot Contamination in Massive MIMO-HBF Systems

In this paper, a novel three-stage data-aided dual-space (DADS) (i.e., beamspace and signal subspace) channel estimation scheme is proposed for massive multiple-input multiple-output hybrid beamforming (m-MIMO-HBF) systems subject to pilot contamination. By exploiting the orthogonality of signal subspace, the non-overlapping interference caused by pilot contamination is identified and mitigated in the coarse channel estimate via subspace projection. Thanks to the independence of transmitted data between users, the overlapping interference is suppressed through alternating iterative refinement of the channel estimate and detected data. Additionally, to initially address the under-determined estimation problem arisen from hybrid beamforming (HBF) structures, an improved matching pursuit algorithm is proposed for coarse sparse beamspace channel estimation by appropriately selecting the scaling factor and adjusting the step size in a piecewise manner, followed by enhancement via subspace projection. Furthermore, by accurately detecting the overlap level of interference in the beamspace, the proposed channel estimation scheme selects an appropriate channel enhancement or refinement strategy to address both non-overlapping and overlapping interference subject to several typical channels without significantly increasing computational complexity. Simulation results demonstrate that the proposed channel estimation scheme achieves higher channel estimation accuracy and exhibits stronger resilience to both interference intensity and the number of interference compared to existing channel estimation schemes.