FedShelter: Efficient privacy-preserving federated learning with poisoning resistance for resource-constrained IoT network

Abstract

Federated Learning (FL) enables decentralized IoT devices to collaboratively train a global model, but its distributed nature makes it vulnerable to poisoning and inference attacks, threatening security and privacy. Previous approaches combine Robust Defense Mechanisms with cryptographic techniques for Robust Secure Aggregation (SecAgg) to ensure security and privacy. However, schemes with strong resistance (robustness) against poisoning attacks often incur high complexity, increasing ciphertext communication and computational overhead, reducing efficiency, and limiting scalability for resource-constrained IoT networks. On the other hand, low-complexity schemes offer higher efficiency but weaker robustness, failing to counter stealthy poisoning attacks like backdoors. To address this, we propose FedShelter, an efficient privacy-preserving FL framework with poisoning resistance for IoT scenarios. It achieves lightweight Robust SecAgg utilizing Secure Two-party Computation (2PC) and incorporates customized encoding techniques to reduce communication overhead and defend against various poisoning attacks. Compared to state-of-the-art solutions like FLAME (USENIX Security’22) and RoFL (S&P’23), FedShelter offers effective robustness against poisoning attacks while reducing communication by up to 56$\times$ and run-time by up to 37$\times$, providing a fast and trustworthy training environment for distributed devices under resource-constrained IoT network.