Learning Linear MMSE Precoder for Uplink Massive MIMO Systems With One-Bit ADCs

Abstract

In this letter, we present a deep learning based linear precoder design to improve the performance for uplink multiuser multiple-input multiple-output (MU-MIMO) systems with one-bit analog-to-digital converters (ADCs). Specifically, by utilizing the Bussgang decomposition based linear minimum mean squared error (MMSE) receiver at the base station (BS), we aim to further minimize the system MSE by optimizing the precoder at each user equipment (UE). To address this problem efficiently, we propose a deep neural network based approach by unfolding the developed projected gradient descent (PGD) algorithm and treating the step size in each PGD iteration as trainable network parameters. Simulation results show that the proposed network design outperforms the conventional PGD algorithm under both perfect and estimated channel state information (CSI) and the improvement is achieved with reduced online complexity.