self-adaptive traffic anomaly detection system for I to T smart home environment

Within Internet of Things (IoT) smart homes, protection of network data streams against unauthorized access and anomalies is vital. This paper proposes an embedded Self-Adaptive Traffic Anomaly Detection Systems (SATADS) in IoT smart homes. The system relies on effective detection and mitigation of traffic anomalies with respect to IoT smart home networks. It has built into it the characteristics of machine learning techniques, supplemented with a Stacking Classifier model to increase the accuracy of anomaly detection, which avails itself of numerous individual classifiers like Random Forest to evaluate characteristics' importance and thereby enhance performance predictively in addition. especially the system is provided with Explainable AI (XAI) techniques to maintain open-ness of the anomaly-detection process for end-users in terms of actionable information for easy understanding and usability. The performance evaluation was conducted using field data (invade.csv) intended for simulating IoT network traffic with useful results showing acceptable performance, mainly in detecting familiar as well as new traffic anomalies. These contributions by the proposed SATADS include self-adaptive, transparent, and trustworthy anomaly detection systems for smart home scenarios that secure the safety and soundness of IoT networks.

Keywords: IoT, Smart Home Environments, Traffic Anomaly Detection, Self-Adaptive Systems, Stacking Classifier, Random Forest, Feature Importance, Explainable AI (XAI), Machine Learning, Cybersecurity, Anomaly Detection System, IoT Security, Transparent AI, Predictive Modeling.