Trust at the Edge: ABAC-Secured Federated Learning for Smart Home Access Control Using Blockchain
Sohanur Rahman, Yi Wang, Bingyang Wei · IEEE Access · 2025
This paper proposes a novel security framework that integrates Attribute-Based Access Control (ABAC) with Federated Learning (FL) using Hyperledger Fabric smart contracts to secure IoT-based smart home environments. With projections estimating over 38 billion IoT connections by 2030 [1], the need for scalable and context-aware access control mechanisms is critical. Our approach enforces fine-grained, attribute-driven policies that govern both user access and device participation in collaborative model training—without compromising data privacy. By deploying our system on a simulated, resource-constrained smart home testbed, we demonstrate that the proposed framework reduces the poisoning attack surface by preventing unauthorized/stale clients and unauthorized access while maintaining computational efficiency. Experimental results show that our method enhances security, preserves the privacy guarantees of FL, and remains feasible for real-world deployment in decentralized smart environments.