Multiclassification of Normal, Abnormal and Encrypted Traffic of IoT

IoT network safety is tested by the development of IoT gadgets and encryption advances. Naive Bayes, C4.5, AdaBoost, and Random Forest have downsides including basic data processing and low multiclassification accuracy. Another multiclassification DL model, the “cost matrix time-space neural network” (CMTSNN), addresses these difficulties. Exploratory outcomes utilizing ToN-IoT and BoT-IoT datasets showed enhancements in accuracy, precision, recall, F1 Score, and false alarm rate contrasted with past methodologies. “Logistic Regression + Gradient Boosting, Naive Bayes + Random Forest, and Stacking Classifier (RF + MLP with LightGBM)” were likewise used to further develop execution. The CMTSNN model detected anomalous encrypted traffic the vast majority of the time.