AI-Driven Intrusion Detection in Internet of Things Networks Using the Edge-IIoTset Dataset

Network Intrusion Detection System (NIDS) is an essential tool in securing cyberspace from a variety of security risks and unknown cyberattacks. A number of solutions have been implemented for Machine Learning (ML), and Deep Learning (DL) based NIDS. In this paper, a deep learning-based multi-class intrusion detection model is introduced on the Edge-IIoTset dataset that includes realistic IIoT network traffic and attack variants. The dataset was preprocessed through feature refinement, categorical encoding, robust scaling, and imbalance handling to enhance model reliability. Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) were implemented and evaluated in terms of stratified 5-fold cross-validation. Experiments prove that the suggested LSTM model is much more effective than CNN and some traditional machine learning methods, and it has an accuracy of 99.96 with close-to-perfect precision, recall, and F1-score. The analyses of confusion matrix and ROC curve also reinforce the high discriminatory power of the model in all attack classes, such as DDoS, Injection, Malware, and Information Gathering. The results indicate the success of sequential deep learning to model intricate traffic patterns in IIoT systems. The study provides a very dependable and scalable intrusion detection system that can be applicable in IIoT cybersecurity in practice.