The rapid proliferation of Internet of Things (IoT) devices across critical domains—such as healthcare, industrial automation, and smart cities—has brought with it a new spectrum of cybersecurity challenges. These devices, often characterized by limited computational capabilities and poor security configurations, are increasingly targeted by sophisticated cyber threats. Traditional intrusion detection systems are not equipped to handle the dynamic, large-scale, and heterogeneous nature of IoT networks, especially under real-time constraints. This paper addresses this critical gap by proposing an AI-based anomaly detection framework tailored specifically for real-time threat mitigation in IoT environments. The primary objective of this study is to develop and evaluate a lightweight, intelligent system capable of detecting anomalous behavior in IoT traffic with high accuracy and minimal latency. The proposed framework leverages machine learning algorithms to model normal device behavior and identify deviations that may indicate malicious activity. Key components include real-time data acquisition, feature extraction, anomaly classification, and automated response mechanisms. Experimental results demonstrate the system’s effectiveness in identifying various categories of cyber threats—including denial-of-service attacks and unauthorized access attempts—with a high detection rate and low false alarm ratio. Furthermore, the implementation is optimized for deployment on edge devices, ensuring scalability and reduced reliance on cloud infrastructure. The findings underscore the potential of real-time AI-driven anomaly detection as a viable and scalable solution for enhancing the resilience of IoT networks against evolving cybersecurity threats.

Keywords: IoT Security, Anomaly Detection, Real-Time Systems, Machine Learning, Cyber Threat Mitigation, Edge Computing