Secure industrial IoT edge computing: a novel intrusion detection paradigm with deep learning

The widespread deployment of Industrial Internet of Things (IIoT) applications has exposed widespread cybersecurity weaknesses, calling for the need for strong Intrusion Detection Systems (IDS). The Edge-IIoTset dataset, a large and realistic dataset for IoT/IIoT application cybersecurity, is employed in this study to investigate the effectiveness of machine learningbased intrusion detection systems. Man-in-the-middle, information gathering, Denial-of-Service (DoS)/Distributed Denial-of- Service (DDoS), injection, and malware are the five principal threat types that are depicted in the Edge-IIoTset, where traffic comes from a number of different IoT devices. The work of this study is suggesting a hybrid deep learning model where gated recurrent units (GRU) are incorporated with convolutional neural networks (CNN). The model is compared with the most sophisticated machine learning techniques currently available, i.e., logistic regression, AdaBoost, random forests, and decision trees. Many experiments reveal that the presented hybrid model is superior to others and shows appreciable results with respect to F1-score, accuracy, precision, and recall. The outcome proves that the suggested model is a great choice for enhancing cybersecurity in IIoT environments because it performs well in classifying intrusions with high accuracy.