Using analog beamforming in mmWave frequency bands we can focus the energy towards a receiver to achieve high throughput. However, this requires the network to quickly find the best downlink beam configuration in the face of non-IID data. We propose a personalized Federated Learning (FL) method to address this challenge, where we learn a mapping between uplink Sub-6GHz channel estimates and the best downlink beam in heterogeneous scenarios with non-IID characteristics. We also devise FedLion, a FL implementation of the Lion optimization algorithm. Our approach reduces the signaling overhead and provides superior performance, up to 33.6% higher accuracy than a single FL model and 6% higher than a local model.
@article{arxiv.2310.00406,
title = {mmWave Beam Selection in Analog Beamforming Using Personalized Federated Learning},
author = {Martin Isaksson and Filippo Vannella and David Sandberg and Rickard Cöster},
journal= {arXiv preprint arXiv:2310.00406},
year = {2023}
}
Comments
10 pages, to be published in IEEE Future Networks World Forum 2023