BeamSeek: Deep Learning-based DOA Estimation for Low-Complexity mmWave Phased Arrays

TL;DR

BeamSeek introduces a deep learning-enhanced method for fast and accurate DOA estimation in low-complexity mmWave phased arrays, outperforming traditional correlation-based techniques especially in noisy environments.

Contribution

It combines agile beam switching with deep learning and data augmentation to improve DOA estimation accuracy and robustness in practical mmWave systems.

Findings

Achieves up to 8-degree reduction in estimation error.

Demonstrates robustness in low SNR conditions.

Validated at 60 GHz using NSF PAWR COSMOS testbed.

Abstract

A novel approach combining agile beam switching with deep learning to enhance the speed and accuracy of Direction of Arrival (DOA) estimation for millimeter-wave (mmWave) phased array systems with low-complexity hardware implementations is proposed and evaluated. Traditional DOA methods requiring direct access to individual antenna elements are impractical for analog or hybrid beamforming systems prevalent in modern mmWave implementations. Recent agile beam switching techniques have demonstrated rapid DOA estimation, but their accuracy and robustness can be further improved via deep learning. BeamSeek addresses these limitations by employing a Multi-Layer Perceptron (MLP) and specialized data augmentation that emulates real-world propagation conditions. The proposed approach was experimentally validated at 60 GHz using the NSF PAWR COSMOS testbed, demonstrating significant improvements over a correlation-based method across various Signal-to-Noise Ratio (SNR) levels. Results show that BeamSeek achieves up to an 8 degree reduction in average estimation error compared to this baseline, with particular advantages in noisy channels. This makes it especially suitable for practical mmWave deployments in environments characterized by multipath interference and hardware constraints.