Angular-Based 3D Hybrid Precoding for URA in Multi-User Massive MIMO Systems

This paper proposes a new angular-based 3D two- stage hybrid precoding scheme for multi-user massive MIMO systems using uniform rectangular arrays (URA), where users are partitioned into different groups based on the similarity of their angle-of-departure (AoD) information. At first using the user-group AoD ranges, the RF- beamforming stage is designed to reduce the inter- group interference, the number of RF chains, and the channel state information (CSI) overhead. Then, the digital baseband precoder stage is constructed via regularized zero-forcing (RZF) technique using the effective channel seen from baseband to reduce the intra-group interference between the users, considering three approaches: joint-group-processing (JGP), per-group-processing (PGP) and common-group-processing (CGP). Illustrative results indicate that the proposed two-stage hybrid precoding schemes with the reduced hardware cost/complexity and relaxed CSI estimation overhead can closely approach the sum- rate performance of the ideal single-stage fully- digital precoding. Moreover, their performance gap becomes negligible as the array size increases.