Federated Learning-driven Beam Management in LEO 6G Non-Terrestrial Networks
2026-03-11 • Machine Learning
Machine Learning
AI summaryⓘ
The authors studied how to choose the best communication beams for satellites in Low Earth Orbit, which is tricky because conditions change a lot. They used a technique called Federated Learning, where different satellite groups learn together without sharing raw data, helped by high-altitude platforms. They tested two computer models, a Multi-Layer Perceptron and a Graph Neural Network, using real beam and signal data. Their results showed the Graph Neural Network predicted the best beams more accurately and steadily, especially when satellites are low in the sky. This work suggests smarter and lighter ways to manage satellite beams in future networks.
Low Earth Orbit (LEO)Non-Terrestrial Networks (NTNs)Beam ManagementFederated LearningHigh-Altitude Platform Stations (HAPS)Multi-Layer Perceptron (MLP)Graph Neural Network (GNN)BeamformingElevation Angle
Authors
Maria Lamprini Bartsioka, Ioannis A. Bartsiokas, Athanasios D. Panagopoulos, Dimitra I. Kaklamani, Iakovos S. Venieris
Abstract
Low Earth Orbit (LEO) Non-Terrestrial Networks (NTNs) require efficient beam management under dynamic propagation conditions. This work investigates Federated Learning (FL)-based beam selection in LEO satellite constellations, where orbital planes operate as distributed learners through the utilization of High-Altitude Platform Stations (HAPS). Two models, a Multi-Layer Perceptron (MLP) and a Graph Neural Network (GNN), are evaluated using realistic channel and beamforming data. Results demonstrate that GNN surpasses MLP in beam prediction accuracy and stability, particularly at low elevation angles, enabling lightweight and intelligent beam management for future NTN deployments.