CSI Feedback Overhead Reduction for 5G Massive MIMO Systems

Abstract

Enhancing the throughput of multi-user (MU) massive multiple-input multiple-output (MIMO) networks is one of the biggest promises that the fifth generation (5G) networks are expected to deliver. In the Third Generation Partnership Project (3GPP) New Radio (NR) standardization efforts, downlink precoding designs that balance performance and uplink feedback overhead are being investigated. Most recently, a high-resolution precoder (Type-II codebook) was specified for downlink NR Release (Rel.) 15 wherein the channel state information (CSI) feedback is compressed in the spatial domain via exploiting a Discrete Fourier Transform (DFT)-based codebook structure. An extension of the Type-II codebook for NR Rel. 16 which also exploits frequency correlation to reduce CSI feedback overhead is currently under study. In this paper, an overview of some of the recent developments for Rel. 16 Type-II codebook is provided. In addition, a practical approach is proposed that uses multi-stage quantization of codebook parameters with variable quantization resolution, where the resolution is proportional to the coefficients' amplitude values. This approach helps provide better utilization of the CSI feedback, compared with the case with the same quantization resolution for all coefficients. System-level simulation results are provided which show that the proposed approach significantly reduces the CSI feedback overhead without notable impact on performance.