A Hybrid Intrusion Detection System for Smart Home Security

Abstract

Although smart home systems contribute to improving many societal challenges, they usually have limited hardware which makes them vulnerable to attack. A security system in a smart home should have the capability to detect any existing or upcoming threats from the network, devices, sensors, or users. In this paper, we use machine learning algorithms for intrusion detection. In particular, we explore a publicly available dataset, CSE-CIC-IDS2018, to discover anomalous activities that can occur in a smart home environment by using a two-tiered system. We use the dataset2018, and run multiple machine learning classification models such as the random forest, xgboost, and decision tree. The algorithms are trained on the decision layer, and our experiments show that each of the models achieve a different level of accuracy in identifying possible anomalies that indicate attacks. The accuracy level achieved by our approach is promising for its practical implementation in a smart home system.