FedOrbit: Adaptive Personalized Federated Learning for Non-IID LEO Satellite Constellations
2026-08-10 • Machine Learning
Machine Learning
AI summaryⓘ
The authors study how to improve federated learning for satellites orbiting Earth, where data is uneven and communication with ground stations is inconsistent. They propose FedOrbit, a method that trains satellites together in a way that accounts for differences in data across orbits and uses smart ways to combine information. Their approach performs better than existing methods on several remote sensing tests, showing more accurate results and more consistent performance across different orbits. This helps handle the challenges caused by satellite movement and varying data distributions.
federated learningLEO satellitesnon-IID dataremote sensinginter-satellite linkshierarchical aggregationclass distributionDirichlet partitioningpersonalizationorbit geometry
Authors
Satwat Bashir, Tasos Dagiuklas, Muddesar Iqbal
Abstract
Federated learning (FL) in Low Earth Orbit (LEO) satellite constellations is affected by non-IID data and irregular ground-station visibility, both driven by orbital geometry. Global aggregation performs poorly when orbit-level class distributions are disjoint, while strong personalisation can be excessive when these distributions overlap. We present FedOrbit, which combines continuous orbit-level training over inter-satellite links, class-aware hierarchical aggregation, quality-weighted feature aggregation with return-rate dampening, and adaptive feature decomposition based on inter-orbit class similarity. Across three remote-sensing benchmarks and two non-IID partitions, FedOrbit achieves the highest accuracy in five of six settings and is within $0.9$ percentage points of the best result in the sixth. The gains over the strongest baseline reach $16.1$ percentage points under Dirichlet partitioning and $8.6$ under pathological partitioning, with the smallest per-orbit accuracy spread in five of six settings.