Hybrid Beamforming/Combining for Millimeter Wave MIMO: A Machine Learning Approach

TL;DR

This paper introduces a deep neural network-based hybrid beamforming method for multi-user millimeter wave MIMO systems, leveraging end-to-end learning to improve performance over traditional techniques.

Contribution

It proposes a novel deep learning framework, DNHB, formulated as an autoencoder, for efficient hybrid beamforming in multi-user mmWave MIMO systems, addressing non-convex optimization challenges.

Findings

DNHB outperforms traditional methods by about 2 dB in BER.

Deep neural networks provide superior representation for beamforming design.

End-to-end self-supervised learning enhances system performance.

Abstract

Hybrid beamforming (HB) has emerged as a promising technology to support ultra high transmission capacity and with low complexity for Millimeter Wave (mmWave) multiple-input and multiple-output (MIMO) system. However, the design of digital and analog beamformer is a challenge task with non-convex optimization, especially for the multi-user scenario. Recently, the blooming of deep learning research provides a new vision for the signal processing of communication system. In this work, we propose a deep neural network based HB for the multi-User mmWave massive MIMO system, referred as DNHB. The HB system is formulated as an autoencoder neural network, which is trained in a style of end-to-end self-supervised learning. With the strong representation capability of deep neural network, the proposed DNHB exhibits superior performance than the traditional linear processing methods. According to the simulation results, DNHB outperforms about 2 dB in terms of bit error rate (BER) performance compared with existing methods.