Smart IoT applications of multi attack detection using cluster F1MI approach

Security infrastructure for Internet of Things (IoT) networks is still a major concern today, as they are constantly violated by cyber attacks. Machine Learning (ML)can efficiently scale to identify abnormal activity in large volumes of IoT traffic. The proposed Smart Secured IoT Framework (SSIF) is used to improve the security of smart IoT networks with the assist of the Bot-IoT dataset. The framework begins with data preprocessing to ensure high-quality input, followed by Cluster F1-MI feature engineering, which uses correlation-based F1-MI to select the most informative attributes and produce a robust, compact feature set. Cluster-based feature validation groups features that are related to cluster the features that are correlated with each other so that a representative feature is selected from the cluster and the correlation is removed to improve efficiency. Our framework uses a variety of ML classifiers, such as SVM, Random Forest, Gradient Boosting, XGBoost, and Neural Networks, for threat classification that is adaptable and accurate. The effectiveness of the classifiers can be tested using metrics like Accuracy, Precision, Recall, F1 Score, and AUC-PR. The random forest can classify attack types with accuracy of more than 0.97. The system also generates an email notification to the admin and activates the alarm when an anomaly is detected. The proposed SSIF proves to be an effective, cheap, and easy solution to secure smart IoT environments from Bot-IoT.