With the proliferation of Internet of Things (IoT) applications in Industry, Agriculture, Healthcare, Smart Cities, Energy Systems, and Education, the number of IoT users is significantly increasing, and consequently, the number of vulnerabilities associated with IoT networks. The emergence of Federated Learning (FL) gives a new direction to improve the security of IoT networks with data privacy. IoT networks generate a huge amount of personal and sensitive data, whose analysis must not compromise data privacy. However, traditional Machine Learning, and Deep Learning methods often fail to ensure data privacy and security during analysis. Federated Learning techniques can preserve data privacy and security with decentralized learning. In this paper, we perform a systematic survey on the recent trends, methods, systems, and frameworks of Federated Learning. We also discuss the significance of Federated Learning in securing IoT networks, threats to Federated Learning, and defense mechanisms incorporating an exhaustive taxonomy on FL. The paper includes possible attack surfaces on IoT networks and the corresponding countermeasures through FL techniques. In addition, we discuss various existing FL frameworks applied in IoT security. Finally, we highlight a list of research issues, challenges, and recommendations for the reader.
Takhellambam et al. (Fri,) studied this question.