An Adaptive AI-Driven Cyber Threat Detection Framework for Securing Heterogeneous IoT Networks

This work proposes an intelligent cybersecurity system built upon Artificial Intelligence (AI) to address evolving cyber threats in heterogeneous Internet of Things (IoT) environments. The proposed framework integrates machine learning with mathematical threat analysis to shift from traditional system security, which responds after an attack, to a proactive approach that predicts and prevents threats. It reacts immediately, processes in just 0.35 s, adapts to 95% of IoT surroundings, and handles security by categorizing threats into four tiers with minimal impact on performance. Tests against standard Intrusion Detection Systems (IDSs), such as SNORT, Suricata, and Bro/Zeek, demonstrate that the framework is superior at handling a wide range of threats.