An Efficient Malware Detection in IIoT Using Machine Learning Techniques

The proliferation of Industrial Internet of Things (IIoT) devices has revolutionized industrial processes, but it has also introduced significant security vulnerabilities. Malware targeting IIoT devices threatens critical infrastructures, requiring efficient, rapid detection systems suitable for resource-constrained environments. This study focuses on achieving extreme resource efficiency in IIoT malware detection by proposing a compact Light Gradient-Boosting Machine (LightGBM) based framework that maintains sub-second inference time and a model size under 1 MB. The framework is evaluated using the dataset called “Edge-IIoTset” and benchmarked against the “UNSW-NB15” dataset. The model achieves high accuracy (98.52<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>), an F1-score of 98.5%, and an inference time of 0.4 seconds, with a model size of 0.7 MB. This demonstrates its capability to deliver robust detection while maintaining operational feasibility for edge devices. Key contributions include the optimization of a lightweight LightGBM model tailored for resource-constrained IIoT environments and a comprehensive evaluation of its performance across datasets. The findings highlight the potential of lightweight, efficient models to enhance IIoT security without compromising detection capabilities.