DEVEOPMENT OF A DEEP LEARNING BASED FRAMEWORK FOR SECURITY OF IOT NETWORK PROTOCOL

This study presents the development of a Feed-Forward Neural Network (FFNN)-based model for security of Internet of Things (IoT) network protocols. The proposed method applied in the execution of the study involves data collection, preprocessing, feature selection, and model training using the CIC-IoT 2022 dataset, which includes normal and attack traffic from various IoT devices. In the study, Synthetic Minority Oversampling Technique (SMOTE) was used for data balancing, Principal Component Analysis (PCA) technique was used for feature selection and hyperparameter optimization were employed to enhance the performance of the model during training. The system was simulated in the NS-3 environment to replicate real-world IoT network conditions, and its effectiveness was evaluated using metrics such as accuracy, precision, recall, and F1-score. The results demonstrated that the FFNN-based model achieves an average validation accuracy of 89.7%, precision of 88.9%, recall of 86.9%, and an F1-score of 87.9%. The system results showcased robustness in detecting various attacks, including DoS, brute force and RTSP attacks in mixed traffic scenarios, meanwhile this study serves as a strong foundation for leveraging deep learning techniques to enhance IoT network security protocols.