Near-field Beam training for Extremely Large-scale MIMO Based on Deep Learning

TL;DR

This paper introduces a deep learning-based near-field beam training method for ELAA systems that reduces training overhead and enhances beamforming performance by leveraging CNNs to learn channel features from historical data.

Contribution

It proposes a novel CNN-based approach for near-field beam training that eliminates the need for predefined codebooks and reduces overhead in ELAA systems.

Findings

Achieves more stable beamforming gain.

Significantly improves performance over traditional methods.

Reduces near-field beam training overhead.

Abstract

Extremely Large-scale Array (ELAA) is considered a frontier technology for future communication systems, pivotal in improving wireless systems' rate and spectral efficiency. As ELAA employs a multitude of antennas operating at higher frequencies, users are typically situated in the near-field region where the spherical wavefront propagates. The near-field beam training in ELAA requires both angle and distance information, which inevitably leads to a significant increase in the beam training overhead. To address this problem, we propose a near-field beam training method based on deep learning. We use a convolutional neural network (CNN) to efficiently learn channel characteristics from historical data by strategically selecting padding and kernel sizes. The negative value of the user average achievable rate is utilized as the loss function to optimize the beamformer. This method maximizes multi-user networks' achievable rate without predefined beam codebooks. Upon deployment, the model requires solely the pre-estimated channel state information (CSI) to derive the optimal beamforming vector. The simulation results demonstrate that the proposed scheme achieves a more stable beamforming gain and significantly improves performance compared to the traditional beam training method. Furthermore, owing to the inherent traits of deep learning methodologies, this approach substantially diminishes the near-field beam training overhead.