Federated Learning for Hybrid Beamforming in mm-Wave Massive MIMO

TL;DR

This paper proposes a federated learning framework for hybrid beamforming in mm-Wave massive MIMO systems, reducing communication overhead and enhancing robustness against data imperfections compared to centralized methods.

Contribution

It introduces a novel FL-based approach for hybrid beamforming that trains models at the base station using user gradients, improving privacy and efficiency.

Findings

FL reduces transmission overhead compared to CML.

The proposed CNN-based method is more tolerant to channel data imperfections.

Numerical results demonstrate improved robustness and efficiency.

Abstract

Machine learning for hybrid beamforming has been extensively studied by using centralized machine learning (CML) techniques, which require the training of a global model with a large dataset collected from the users. However, the transmission of the whole dataset between the users and the base station (BS) is computationally prohibitive due to limited communication bandwidth and privacy concerns. In this work, we introduce a federated learning (FL) based framework for hybrid beamforming, where the model training is performed at the BS by collecting only the gradients from the users. We design a convolutional neural network, in which the input is the channel data, yielding the analog beamformers at the output. Via numerical simulations, FL is demonstrated to be more tolerant to the imperfections and corruptions in the channel data as well as having less transmission overhead than CML.