<![CDATA[The development of the Industrial Internet of Things (IIoT) has increased the efficiency and productivity of industrial systems, but has simultaneously increased vulnerability to cyber threats, particularly malware attacks that can disrupt computer systems, networks, and IoT devices. This study aims to build and evaluate a machine learning model for classifying malware threats in IIoT infrastructure. The dataset used is the CIC IIoT Dataset 2025, containing normal (benign) network traffic and two types of malware attacks, namely Mirai SYN Flood and Mirai UDP Flood. The main problem in the dataset is class imbalance, so the SMOTE technique is applied to balance the data distribution, alongside the challenge of similar characteristics between attack classes. Three machine learning algorithms, Support Vector Machine (SVM), K-Nearest Neighbors (KNN), and Random Forest, were used to build multi-class and binary classification models. Model performance was evaluated using accuracy, precision, recall, F1-score, confusion matrix, and ROC-AUC curves. The multi-class results show that Random Forest achieved the best performance with 87% accuracy and a weighted F1-score of 0.87, followed by SVM (86%; 0.86) and KNN (85%; 0.86), while in the binary classification scenario Random Forest reached 98.42% accuracy. These results are expected to provide insight into the most effective algorithm for detecting and classifying malware attacks in IIoT networks, thereby improving the security and reliability of IIoT-based industrial systems.]]>
Copyrights © 2026
