A Comprehensive Review on Security of Internet of Things using Machine Learning: Threats, Recent Advancements and Challenges

Abstract

As technology has advanced, Internet of Things (IoT) devices have become more prevalent in daily life. However, as more people use IoT devices, their security becomes a bigger concern for both manufacturers and users. Despite considerable efforts from the researchers, improving the detection accuracy while minimizing false positives and effectively identifying new types of security breaches remain critical issues. Recently, techniques such as deep learning and machine learning have been explored as potential solutions to improve the security of Internet of Things devices. This study offers an approach for safeguarding IoT environments based on major ML and DL techniques. It categorizes selected studies according to the specific ML/DL algorithms applied. Additionally, this review comprehensively examines the latest advancements in ML and DL approaches by detailing their methodologies, evaluation metrics, and choice of datasets. By identifying the limitations of current methods, this work outlines research challenges and suggestions for future investigation to improve the security of IoT devices.