Integrating Statistical Methods and Game Theory for Enhanced IoT Intrusion Detection

Intrusion Detection Systems (IDS) are widely used to secure networks and detect malicious traffic. The proliferation of IoT devices has heightened the focus on protecting these vulnerable devices. Modern IDSs leverage AI to enhance attack detection, but the complexity of network traffic poses challenges in identifying attack traits, which are crucial for reliable and confident classification. In light of these considerations, this work introduces a new comprehensive framework for analyzing datasets, first reducing features using statistical methods like the Pearson Correlation Coefficient, and then applying the Shapley Value from game theory to better understand attacks. This approach standardizes dataset analysis and enhances the reproducibility of machine learning experiments, addressing a key challenge in AI-based IDS research.