Multiuser Joint Downlink Channel Reconstruction Based On Spatial Consistency

Massive multiple-input multiple-output (MIMO) systems working in frequency division duplex (FDD) transmission mode are facing great challenges because of the heavy training and feedback overhead. In this paper, we propose a novel algorithm for joint downlink channel reconstruction of multiple close-by users with reduced complexity and feedback overhead. Based on the spatial consistency, we extend the orthogonal matching pursuit algorithm to the multiuser case to jointly estimate the common channel parameters of these users. Specifically, a hierarchical dictionary is introduced to iteratively identify the common paths with reduced computational complexity. On another hand, based on the spatial reciprocity, the common path parameters are frequency-independent and can be obtained in the uplink, and only frequency-dependent channel parameters are then estimated in the downlink with reduced downlink training and feedback overhead. Numerical results verify the effectiveness and robustness of the proposed algorithm, and demonstrate that the proposed scheme can reduce the feedback overhead of channel estimation of adjacent users. (C) 2021 Elsevier B.V. All rights reserved.