Channel Estimation for Massive MU-MIMO Systems with Real Image Denoising Network

Abstract

Channel estimation is one of the critical challenges for massive multiuser multiple-input multiple-output (MU-MIMO) systems. In this paper, a deep learning (DL) method, exploiting the sparsity of the massive MIMO channel, is proposed to improve the performance of least squares (LS) estimation. Specifically, we first consider the sparse massive MIMO channel matrix as a natural image. Then, a novel channel estimation method based on real image denoising network (RIDNet) is proposed to effectively mitigate the impact of noise on LS estimation. Finally, simulation results are provided to corroborate the superiority of the proposed method in performance and robustness.