3D Angular-Based Hybrid Precoding for Multi-Cell MU-Massive-MIMO Systems in C-RAN Architecture

This paper proposes a three-dimensional angular-based hybrid precoding (3D-AB-HP) for multi-cell multi-user massive multiple-input multiple-output (MU-mMIMO) systems. By employing the angle-of-departure (AoD) information, the RF precoder in the proposed 3D-AB-HP is designed to decrease the channel estimation overhead and the number of RF chains. Then, the baseband (BB) precoder in the 3D-AB-HP is constructed via regularized zero-forcing technique using the effective channel seen at the BB. To further mitigate the inter-cell interference for the cell-edge users, we propose three cooperation strategies for the multi-cell downlink transmission in the case of cloud radio access networks (C-RAN): (i) non-cooperation (NC), (ii) partial-cooperation (PC) and (iii) full-cooperation (FC). Illustrative results indicate that FC and PC improve remarkably the sum-rate performance via utilization of C-RAN in comparison to the traditional RAN. Moreover, by means of the directional beamforming, the proposed 3D-AB-HP with PC and NC can outperform the corresponding single-stage fully-digital precoding requiring higher hardware cost/complexity and larger channel estimation overhead.