A Novel Codebook Based CSI Reconstruction Algorithm in FDD Massive MIMO Systems

Abstract

Channel state information (CSI) plays a critical role for beamforming technology in frequency division duplex (FDD) massive multiple-input and multiple-output (MIMO) systems. In order to improve the beamforming accuracy, we present a novel codebook based CSI reconstruction algorithm to enhance CSI feedback performance, thereby improving communication quality. Based on the existing codebook feedback mechanism, initially, we leverage channel temporal correlation to smooth the precoding matrix using a sliding window approach, where both Type I and Type II codebooks are considered. Subsequently, a minimum mean square error (MMSE) based modification technique is introduced to modify the reconstruction result. The proposed CSI reconstruction method incorporates an adjustable weighting parameter within the sliding window, rendering it adaptable to channels with different user speeds, such as high-speed train communication channel. Moreover, our algorithm entirely operates at the base station (BS). Therefore, there is no additional feedback overhead compared with traditional codebook based CSI feedback methods. Numerical results present the proposed algorithm outperforms baselines for beamforming.