Intelligent Control Agent For IoT Network Attack Detection: A Machine Learning Approach

This investigation addresses critical security vulnerabilities in Internet of Things (IoT) architectures through the implementation of a machine learning-enhanced subsystem within the Internet of Things intelligent control agent (IoT-ICA) framework. The methodology integrates four distinct algorithmic approaches—support vector machines, random forests (RFs), XGBoost, and light gradient boosting machine (LGBM)—evaluated against the MedBotIoT dataset’s heterogeneous device configurations. Empirical analysis demonstrates LGBM’s superior efficacy, achieving 99.04% threat detection rates and 99.02% classification accuracy. The implementation’s computational efficiency is optimized through strategic feature selection protocols while maintaining high-precision threat detection capabilities. This research advances IoT security through two principal contributions: the development of an adaptive monitoring system and comprehensive algorithmic performance evaluation in resource-constrained environments, establishing foundational methodologies for robust IoT security implementations.