Machine Learning Techniques for IoT Security: A Bibliometric Analysis and Road-Mapping

Abstract

The integration of the Internet of Things (IoT) across various sectors, including healthcare, smart cities, and industrial automation, has heightened the demand for robust security solutions to counter the threat of cyberattacks. Cyberattacks targeting IoT devices can have significant consequences, including data breaches, operational disruptions, and substantial network compromises. Machine learning (ML) has become an essential approach for addressing these complex security challenges. Therefore, this study conducts a bibliometric analysis to quantitatively map the intellectual structure and thematic development of the ML for the IoT security research domain. An analysis was conducted on 3,744 scientific documents, published from 2015 to 2025, obtained from the Scopus database. Performance analysis and science mapping techniques, including keyword co-occurrence and co-authorship analyses, were employed to identify key contributors, collaboration networks, organisations, and sources. The results indicate that the field is substantially influenced by dominant AI trends, particularly the rise of large language models (LLMs), resulting in a notable “practicality gap” between computationally intensive models and the resource-constrained nature of IoT devices. Additionally, the country collaboration map indicates that the United States, China, the United Kingdom, and India are the largest contributors to publication volume. In contrast, federated learning has become a significant area of research, focusing on data privacy and decentralisation in IoT. Finally, this study identifies potential research gaps by examining current trends, thereby providing insights for future work in developing efficient and resilient IoT security.