XRFLWID: XGBoost and Random Forest-Based Lightweight Intrusion Detection Model for IoT Attack Detection

Due to the IoT’s rapid growth, it is projected that billions of IoT devices will be linked to the Internet of Things. While IoT security flaws make it simple for attackers to take advantage of the devices and transform them into a botnet, however, systems for detecting intrusions can be quite important in solving IoT security loopholes and enhancing the IoT metasystem’s well-being. Modish machine learning approaches have been succeeding and are mostly used in designing and developing network intrusion detection systems to tackle the modern cyber assaults toward the IoT infrastructure. In this manuscript, we are proposing a lightweight anomaly-based intrusion detection system with novel preprocessing steps. The proposed intrusion detection model uses XGBoost-based feature selection and a random forest classifier for binary classification. The BoT-IoT dataset by New South Wales University Sydney, which contains several attack types incorporating DDoS strike categories, is being used in the perusal. Normal and attack labels of the dataset are heavily imbalanced, and Synthetic Minority Oversampling TEchnique (SMOTE) is used to balance the classes.