Privacy-Preserving Federated Learning for Intrusion Detection in IoT Environments: A Survey
Abhishek Vyas, Po‐Ching Lin, Ren‐Hung Hwang, Meenakshi Tripathi
Abstract
With the rapid development of artificial intelligence and a new generation of network technologies, the Internet of Things (IoT) is expanding worldwide. Malicious agents consistently exploit new technical vulnerabilities to access the various IoT systems used in critical industries, medical diagnosis, military, and defense systems. To mitigate these threats, IoT networks should be equipped with intrusion detection systems capable of detecting threat vectors in an attempt to compromise the systems. Moreover, many researchers have integrated privacy-preserving technologies such as homomorphic encryption, differential privacy, and secure multiparty computation with machine learning algorithms. Furthermore, federated learning, which shares only model parameters rather than data, provides distributed privacy-preserving learning; therefore, federated learning is secure and reliable for the implementation of intrusion detection systems in IoT environments. This survey examined the utilization and applications of privacy-preserving mechanisms, explicitly focusing on privacy-preserving federated learning for intrusion detection systems in IoT environments. This survey also highlights future research directions and open research questions. Privacy-preserving federated learning can significantly contribute to the rapid and efficient detection and prevention of various threat vectors that target IoT ecosystems.