Abstract:

The proliferation of Internet of Things (IoT) devices coupled with the widespread adoption of encryption technology has posed significant challenges to IoT cybersecurity. The surge in encrypted abnormal traffic among IoT devices necessitates robust methods for identifying and mitigating potential threats. Existing detection methods often suffer from limitations such as simplistic data processing, inadequate feature extraction, data imbalance, and low multiclassification accuracy. In response to these challenges, this project aims to propose a novel approach for identifying abnormal encrypted traffic in IoT networks. The primary objective is to develop a multiclassification deep learning model, termed the cost matrix time–space neural network (CMTSNN), tailored specifically for this task. The key focus lies in addressing the shortcomings of existing methods by enhancing feature extraction robustness, handling data imbalance, and improving overall classification accuracy. Experimental evaluations were conducted utilizing datasets including ToN-IoT, BoT-IoT,. Comparative analysis against existing methods demonstrated superior performance across various metrics including accuracy, precision, recall, F1 Score, and false alarm rate. The CMTSNN model exhibited notable improvements in classification accuracy, particularly for minority categories, thereby enhancing the overall multiclassification performance. And also added in the project is voting classifier (RF + AdaBoost + MLP) and CNN-LSTM models, those are employed tom improve the performance , the project attains 99% accuracy in detecting abnormal encrypted traffic. A user-friendly Flask-based front end facilitates easy testing and interaction, while robust user authentication ensures secure access. These enhancements solidify the system's effectiveness in IoT cybersecurity, reinforcing its reliability and usability in realworld applications.

Description:

Abnormal and encrypted traffic classification, cost penalty matrix, deep learning (DL), Internet of Things (IoT).