FusionNet: Enhanced Beam Prediction for mmWave Communications Using Sub-6GHz Channel and A Few Pilots

TL;DR

This paper introduces FusionNet, a neural network that predicts mmWave beamforming using uplink sub-6GHz data and minimal mmWave pilots, enhancing accuracy and robustness especially in low SNR conditions.

Contribution

The paper presents a novel dual-input neural network architecture, FusionNet, that leverages sub-6GHz channels and few pilots for improved mmWave beam prediction, with new data preprocessing techniques.

Findings

FusionNet outperforms existing methods in accuracy.

The approach is robust at low SNR levels.

Data augmentation improves model generalization.

Abstract

In this paper, we propose a new downlink beamforming strategy for mmWave communications using uplink sub-6GHz channel information and a very few mmWave pilots. Specifically, we design a novel dual-input neural network, called FusionNet, to extract and exploit the features from sub-6GHz channel and a few mmWave pilots to accurately predict mmWave beam. To further improve the beamforming performance and avoid over-fitting, we develop two data pre-processing approaches utilizing channel sparsity and data augmentation. The simulation results demonstrate superior performance and robustness of the proposed strategy compared to the existing one that purely relies on the sub-6GHz information, especially in the low signal-to-noise ratio (SNR) regions.