A Novel Quantization Method for Deep Learning-Based Massive MIMO CSI Feedback

Abstract

In massive multiple-input multiple-output (MIMO) systems, channel state information (CSI) needs feeding back to the base station (BS) by user equipment (UE) to attain the potential benefits of massive MIMO. But the large number of antennas at the BS causes a huge feedback overhead, thereby making it prohibitive to realize CSI feedback in massive MIMO. Deep leaning-based (DL) compressive sensing methods for CSI feedback can potentially reduce the overhead significantly. However, without quantization, a data-bearing bitstream for transmission cannot be produced at the UE. In this paper, we propose a novel quantization framework and training strategy for DL-based CSI feedback, which not only makes the current CSI feedback network applicable in real communication systems but also minimizes the introduced quantization distortion to improve the reconstruction quality. Experimental results demonstrate that the proposed quantization method performs well and is robust to quantization errors.