ASA: Adaptive Smart Agent Federated Learning via Device-Aware Clustering for Heterogeneous IoT

2026-02-16

AI Summary

This work addresses the challenges of efficiency and fairness in federated learning across heterogeneous Internet of Things (IoT) devices, which exhibit significant disparities in computation, memory, and network capabilities. To tackle these issues, the authors propose the Adaptive Scheduling and Aggregation (ASA) framework, which employs lightweight intelligent agents to continuously monitor device resources and dynamically cluster them into high-, medium-, and low-capacity groups. ASA integrates this real-time resource-aware clustering with hierarchical, customized model architectures to enable adaptive collaborative training. By synergistically combining dynamic clustering based on live resource profiling with tailored model design, the framework substantially enhances system efficiency and inclusivity of device participation. Experimental results demonstrate that ASA achieves 98.89% and 85.30% accuracy on MNIST and CIFAR-10, respectively, while reducing communication overhead by 43%–50% and improving resource utilization by 43%.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Multiagent Systems: Adversarial Agents
• Application Domains: Internet of Things, Sensor Networks & Smart Cities

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
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Federated learning (FL) has become a promising answer to facilitating privacy-preserving collaborative learning in distributed IoT devices. However, device heterogeneity is a key challenge because IoT networks include devices with very different computational powers, memory availability, and network environments. To this end, we introduce ASA (Adaptive Smart Agent). This new framework clusters devices adaptively based on real-time resource profiles and adapts customized models suited to every cluster's capability. ASA capitalizes on an intelligent agent layer that examines CPU power, available memory, and network environment to categorize devices into three levels: high-performance, mid-tier, and low-capability. Each level is provided with a model tuned to its computational power to ensure inclusive engagement across the network. Experimental evaluation on two benchmark datasets, MNIST and CIFAR-10, proves that ASA decreases communication burden by 43% to 50%, improves resource utilization by 43%, and achieves final model accuracies of 98.89% on MNIST and 85.30% on CIFAR-10. These results highlight ASA's efficacy in enhancing efficiency, scalability, and fairness in heterogeneous FL environments, rendering it a suitable answer for real-world IoT apps.

Problem

• Federated Learning
• Device Heterogeneity
• IoT
• Resource Constraints
• Model Adaptation

Innovation

• Federated Learning
• Device Heterogeneity
• Adaptive Clustering
• Resource-Aware Model Customization
• IoT