Update Estimation and Scheduling for Over-the-Air Federated Learning with Energy Harvesting Devices

2025-01-30

AI Summary

Over-the-air (OTA) federated learning for energy-harvesting devices suffers from convergence bias and communication redundancy due to channel fading, energy constraints, and data heterogeneity. To address these challenges, we propose an entropy-driven dual-paradigm scheduling framework: entropy-optimized scheduling for known data distributions, and—when distributions are unknown—a least-squares-based user representation model enabling adaptive client selection, thereby pioneering the integration of diversity-awareness into energy-efficient collaborative updating. Our framework jointly optimizes OTA aggregation, fine-grained energy harvesting modeling, and fading channel adaptation. Theoretical analysis guarantees convergence, while experiments demonstrate significant improvements over baseline methods: 32% reduction in communication rounds, 27% lower energy consumption, and a 4.8% gain in global model accuracy. Moreover, the approach exhibits enhanced robustness against data skewness and channel fluctuations.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Planning, Routing, and Scheduling: Planning/Scheduling and Learning
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We study over-the-air (OTA) federated learning (FL) for energy harvesting devices with heterogeneous data distribution over wireless fading multiple access channel (MAC). To address the impact of low energy arrivals and data heterogeneity on global learning, we propose user scheduling strategies. Specifically, we develop two approaches: 1) entropy-based scheduling for known data distributions and 2) least-squares-based user representation estimation for scheduling with unknown data distributions at the parameter server. Both methods aim to select diverse users, mitigating bias and enhancing convergence. Numerical and analytical results demonstrate improved learning performance by reducing redundancy and conserving energy.

Problem

• Wireless Federated Learning
• Energy Harvesting
• Time Scheduling

Innovation

• Energy Harvesting Devices
• User Scheduling Strategies
• Wireless Federated Learning