Downlink statistical precoding scheme for spatially correlated massive MIMO systems

Abstract

In this paper, we consider spatial beamforming for massive multiple-input multiple-output (MIMO) systems in frequency division duplexing (FDD). In order to reduce channel state information (CSI) feedback overhead of users, near-by users that have similar statistical channel correlation are put into a group and channel correlation matrix based precoding scheme is proposed. More specifically, we propose to employ two-stage precoder at a base station (BS): Outer precoder and inner precoder. The former is designed to minimize inter-group interference by considering the statistical channel correlation. The latter is designed to minimize intra-group interference by utilizing modified statistical channel correlation and instantaneous channel state information (ICSI). The inner precoder is composed of three steps: precoded pilot signal transmission phase at the BS, effective channel gain calculation and feedback stage at the mobile user, and data transmission stage at the BS.