Machine Learning for Intrusion Detection in IIoT: A Comprehensive Review
Aissétou Ba, Mehdi Add · Procedia Computer Science · 2025
In recent years, the Industrial Internet of Things (IIoT) has grown rapidly, with the potential to transform and connect many industries, opening up major opportunities for the global economy. IIoT extends to industries such as manufacturing, logistics, transportation, energy, oil and gas, mining and aviation. However, despite its advantages, the IIoT is exposed to increasing cyber attacks due to the large amount of data generated by its many sensors. These attacks create an increased need for advanced security solutions. Intrusion detection systems (IDS), which monitor network traffic and identify abnormal behavior, are essential to protect IIoT networks. Machine learning (ML) is an effective solution to improve IDS performance by enabling proactive intrusion detection. This literature review explores ML-based detection techniques, highlighting supervised, unsupervised and deep learning methods. The main objective of this research is to present and analyze different approaches of machine learning applied to intrusion detection in IIoT, as well as the data sets used and comparative results obtained in existing studies. We also highlight the challenges and limitations of these approaches, such as managing resources, reducing false positives and adapting to new threats. This review concludes that the use of supervised or hybrid models and solutions, such as supervised or federated learning, could improve the security of IIoT systems. Finally, this research provides future directions to overcome the challenges of model scalability and optimization in order to better meet the increasing safety requirements in industrial environments.