QuantFL: Sustainable Federated Learning for Edge IoT via Pre-Trained Model Quantisation

2026-03-18

AI Summary

This work addresses the high energy consumption and carbon footprint incurred by frequent model update uploads in federated learning on edge Internet-of-Things (IoT) devices. To mitigate this, the authors propose QuantFL, a novel framework that integrates pretrained model initialization with a bucket quantization strategy to enable lightweight, aggressive quantization without relying on complex error feedback mechanisms. This approach substantially reduces both uplink and downlink communication overhead, facilitating efficient low-bit transmission. Experimental results demonstrate that QuantFL achieves 89.00% accuracy on MNIST and 66.89% on CIFAR-100, while reducing total communication volume by 40% and saving over 80% of uplink bits compared to baseline methods.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Application Domains: Internet of Things, Sensor Networks & Smart Cities
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Federated Learning (FL) enables privacy-preserving intelligence on Internet of Things (IoT) devices but incurs a significant carbon footprint due to the high energy cost of frequent uplink transmission. While pre-trained models are increasingly available on edge devices, their potential to reduce the energy overhead of fine-tuning remains underexplored. In this work, we propose QuantFL, a sustainable FL framework that leverages pre-trained initialisation to enable aggressive, computationally lightweight quantisation. We demonstrate that pre-training naturally concentrates update statistics, allowing us to use memory-efficient bucket quantisation without the energy-intensive overhead of complex error-feedback mechanisms. On MNIST and CIFAR-100, QuantFL reduces total communication by 40\% ($\simeq40\%$ total-bit reduction with full-precision downlink; $\geq80\%$ on uplink or when downlink is quantised) while matching or exceeding uncompressed baselines under strict bandwidth budgets; BU attains 89.00\% (MNIST) and 66.89\% (CIFAR-100) test accuracy with orders of magnitude fewer bits. We also account for uplink and downlink costs and provide ablations on quantisation levels and initialisation. QuantFL delivers a practical, "green" recipe for scalable training on battery-constrained IoT networks.

Problem

• Federated Learning
• Edge IoT
• Energy Efficiency
• Carbon Footprint
• Pre-trained Models

Innovation

• Federated Learning
• Model Quantisation
• Pre-trained Models
• Edge IoT
• Sustainable AI