Systematic Analysis of Federated Learning Approaches for Intrusion Detection in the Internet of Things Environment

The Internet of Things (IoT) has transformed various sectors by connecting devices, services, and applications, which boosts intelligence and operational efficiency. However, this growing connectivity has heightened concerns regarding data privacy and security, particularly for sensitive information. Intrusion detection systems (IDSs) have emerged as an effective strategy through incorporating machine learning (ML) methodologies to identify sophisticated cybersecurity threats. Traditional IDS solutions typically rely on centralized learning architectures, where data from edge nodes are transferred to central servers for analysis. This approach is effective but introduces major risks to the privacy of IoT data. Artificial intelligence (AI), specifically federated learning (FL), is a new frontier that offers an alternative option by enabling direct implementation of distributed ML on IoT devices without requiring sharing of raw data. FL supports decentralized and privacy-preserving learning, which makes it a critical instrument for enhancing the security of IoT ecosystems. Research on data security and privacy in FL is increasingly becoming popular at present. However, the development of FL for IoT remains in its infancy and requires exploration in various areas. Moreover, current survey studies are limited and overlook the vital aspects of FL, particularly the diversity of datasets and the most recent methodologies employed in FL-enabled IDSs. This study conducts a systematic literature review that examines 43 studies published between 2020 and 2024. The focus revolves on recent advancements in applying AI to intrusion detection within IoT environments. This study provides a comprehensive analysis of key ML models and datasets used to develop an IDS tailored for IoT systems. The review also explores emerging trends, challenges, and opportunities in implementing FL-based IDSs in IoT networks. The findings highlight the potential of FL to tackle pressing privacy and security challenges while enabling scalable and adaptive threat detection solutions. The study highlights important research gaps and proposes critical future directions to advance the field.