IIoT Network Intrusion Detection Using Machine Learning
Abdulrahman Mahmoud Eid, Ali Bou Nassif, Bassel Soudan, Mohammad Noor Injadat
Abstract
Industrial Internet of Things (IIoT) systems are subject to intrusion attacks that may cause devastating damage. However, common security mechanisms may not be applicable as IIoT systems and standard IT networks are fundamentally different. Therefore, it is necessary to consider different security factors and performance metrics when addressing the vulnerabilities and security demands of IIoT. Machine Learning (ML) has proven to be an effective tool in the development of security mechanisms and Intrusion Detection Systems (IDS) for IIoT. This paper presents the implementation of multiple IDS models using six ML algorithms. These models were evaluated on the WUSTL-IIoT-2021 dataset, and their performance was analyzed and compared. Among the six models, the model constructed using the Random Forest (RF) algorithm achieved an accuracy of 99.97%, which is significantly higher than the accuracy achieved by previously reported models in the literature. The results demonstrate the significance of this work and highlight the potential of ML-based IDS models in ensuring the security and safety of IIoT systems.