Privacy-Preserving Defense: Intrusion Detection in IoT using Federated Learning
Leonardo Almeida, Pedro Rodrigues, Rafael Bastos Teixeira, Mário Antunes, Rui L. Aguiar · 2024
The Internet of Things (IoT) presents unprecedented challenges in network security due to its vast deployment and resource limitations. Addressing these challenges requires robust Intrusion Detection Systems (IDS) capable of detecting and mitigating potential threats effectively. In this study, we explore the efficacy of Federated Learning (FL) in training IDS models while ensuring data privacy in IoT scenarios. We leverage three state-of-the-art datasets to evaluate FL-based training approaches with varying numbers of workers. Our experiments demonstrate that FL-based training yields comparable performance to traditional single training methods across multiple performance metrics. Additionally, FL training exhibits faster convergence times, highlighting its efficiency and scalability for training IDS models in IoT environments. These findings underscore the potential of FL as a privacy-preserving technique for enhancing network security in IoT deployments.