Energy-Efficient Federated Learning for Edge Real-Time Vision via Joint Data, Computation, and Communication Design

2025-08-03

AI Summary

To address high energy consumption, limited data volume, and non-IID data distributions in edge-based real-time vision applications over unreliable wireless networks, this paper proposes FedDPQ, a novel federated learning framework. FedDPQ jointly optimizes four dimensions: diffusion-based data augmentation, model pruning, gradient quantization, and adaptive transmission power control. It establishes an energy-convergence closed-loop analytical model and employs Bayesian optimization for joint hyperparameter tuning—the first such integration in federated edge learning. Compared to state-of-the-art methods, FedDPQ significantly improves training convergence speed and stability, reduces edge-device energy consumption by 32.7%, and mitigates the impact of communication interruptions. Extensive evaluation on representative vision tasks demonstrates its efficiency and deployability on resource-constrained edge devices.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Security and privacy of machine learning and AI applications

Abstract

Emerging real-time computer vision (CV) applications on wireless edge devices demand energy-efficient and privacy-preserving learning. Federated learning (FL) enables on-device training without raw data sharing, yet remains challenging in resource-constrained environments due to energy-intensive computation and communication, as well as limited and non-i.i.d. local data. We propose FedDPQ, an ultra energy-efficient FL framework for real-time CV over unreliable wireless networks. FedDPQ integrates diffusion-based data augmentation, model pruning, communication quantization, and transmission power control to enhance training efficiency. It expands local datasets using synthetic data, reduces computation through pruning, compresses updates via quantization, and mitigates transmission outages with adaptive power control. We further derive a closed-form energy-convergence model capturing the coupled impact of these components, and develop a Bayesian optimization(BO)-based algorithm to jointly tune data augmentation strategy, pruning ratio, quantization level, and power control. To the best of our knowledge, this is the first work to jointly optimize FL performance from the perspectives of data, computation, and communication under unreliable wireless conditions. Experiments on representative CV tasks show that FedDPQ achieves superior convergence speed and energy efficiency.

Problem

• Energy-efficient FL for edge real-time vision applications
• Addressing resource constraints in FL via joint optimization
• Mitigating unreliable wireless impact on FL performance

Innovation

• Diffusion-based data augmentation expands datasets
• Model pruning reduces computation energy
• Quantization and power control optimize communication