CSI Feedback with Model-Driven Deep Learning of Massive MIMO Systems

TL;DR

This paper introduces a novel model-driven deep learning approach for CSI feedback in massive MIMO systems, significantly reducing feedback overhead and improving efficiency with a two-stage low rank scheme and an adaptive neural network.

Contribution

It proposes a two-stage low rank CSI feedback scheme and a new deep neural network, FISTA-Net, that unfolds an optimization algorithm for improved feedback in massive MIMO systems.

Findings

Outperforms existing algorithms in various scenarios

Reduces feedback overhead effectively

Enhances CSI feedback efficiency with deep learning

Abstract

In order to achieve reliable communication with a high data rate of massive multiple-input multiple-output (MIMO) systems in frequency division duplex (FDD) mode, the estimated channel state information (CSI) at the receiver needs to be fed back to the transmitter. However, the feedback overhead becomes exorbitant with the increasing number of antennas. In this paper, a two stages low rank (TSLR) CSI feedback scheme for millimeter wave (mmWave) massive MIMO systems is proposed to reduce the feedback overhead based on model-driven deep learning. Besides, we design a deep iterative neural network, named FISTA-Net, by unfolding the fast iterative shrinkage thresholding algorithm (FISTA) to achieve more efficient CSI feedback. Moreover, a shrinkage thresholding network (ST-Net) is designed in FISTA-Net based on the attention mechanism, which can choose the threshold adaptively. Simulation results show that the proposed TSLR CSI feedback scheme and FISTA-Net outperform the existing algorithms in various scenarios.