Hybrid Anomaly Detection Framework for Matter-Enabled IoT Devices

Communications reliability and security of Matter-enabled IoT networks are now gaining even more prominence with their widespread acceptance and deployment. Nonetheless, such advancements indeed introduce further challenges, most notably in the detection of the diverse set of anomalies caused by device faults and DDoS attacks that may destabilize the network. This paper presents a hybrid anomaly detection system specifically designed for Matter-enabled IoT environments where rule-based filtering is incorporated with an ML-based detection approach leveraging Extreme Gradient Boosting (XGBoost). The rule-based module detects potential anomalies by applying predefined threshold conditions, and the XGBoost aims to curtail this detection into high detection capability and reduced false positives. The system is experimented with making the prediction on IoT telemetry data, achieving 97% precision, recall, and accuracy-greatly exceeding the lone detection strategies. The results prove how our approach can help mitigate the occurrence of false alarms, alongside imbalanced datasets being major obstacles in IoT networks' anomaly detection. Future research will further include the real-time detection application of sophisticated temporal models, scalable improvements, and security optimization within the confines of resource-constrained Matter-enabled IoT deployment.