Detecting Attacks on Industrial Internet of Things (IIoT) Networks Using the Random Forest Algorithm with Feature Selection and Parameter Optimization Techniques

Abstract

The increasing adoption of interconnected industrial devices has created new cybersecurity challenges in Industrial Internet of Things (IIoT) environments. As communication among sensors, controllers, and monitoring systems becomes more intensive, industrial networks become more vulnerable to cyberattacks that may disrupt operational processes. Existing studies have primarily focused on limited attack categories or conventional IoT environments, creating a need for more effective multi-class attack detection models for complex IIoT networks. To address this issue, this research develops a multiclass intrusion detection framework using Random Forest and Decision Tree classifiers enhanced through feature reduction and hyperparameter optimization techniques. Experiments were conducted using the CICIIoT2025 dataset following data preprocessing, balancing, normalization, and feature selection procedures. The optimized Random Forest model achieved the strongest overall performance across all evaluation metrics. Statistical validation using the Wilcoxon Signed-Rank Test confirmed that the observed improvements were significant. These findings demonstrate that combining feature selection with parameter optimization can strengthen attack detection capabilities in IIoT networks.