Enhancing IOT Security: A review of Machine Learning-Driven Approaches to Cyber Threat Detection

Abstract

Internet Of Things (IOT) is rapidly adopted and implemented across various industries. The fast growth of IOT devices poses a risk, as these devices are ideal targets to be breached and exploited. However, given the heterogeneous nature and resource limitations of IOT networks, the traditional security mechanisms often fail to provide the required security. This study investigates recent IOT security breaches and showcases vulnerabilities exploited by attackers, as well as their impact on consumer, industrial, and healthcare IOT systems. The proposed solutions through ML and DL-driven security are summarized for adaptive threat detection, anomaly-based intrusion prevention, and intelligent risk mitigation. We also analyzed different approaches based on ML and DL to identify and prevent cyber-attacks as an effective solution. These ML and DL – based research papers have been reviewed within the IEEE repository and the publications span from 2020 to 2024, ensuring current literature on IOT security. The results highlight that security models based on ML and DL techniques improve resilience against IOT by allowing real-time detection of attacks, reducing the volume of false positives, and adapting to new threats. Furthermore, this work identifies the existing barriers to the adoption of ML/DL technologies for IOT security and emphasizes the potential areas for future research that may solidify the overall security framework for IOT ecosystems.