A Hybrid Machine Learning Approach to Anomaly Detection in Industrial IoT

Abstract

IIoT is the integration of conventional IoT principles into industrial operations. IIoT has a wide range of practical applications, including but not limited to supply chain management, connected cars, smart grids, smart cities, and smart homes. Regrettably, these systems are increasingly becoming the focus of cybercrime attacks. Machine learning is a promising technology for creating and implementing resilient security measures in IIoT networks. A new and innovative approach to detecting cyberattacks in the IIoT is proposed in this document, through the use of a hybrid machine classifier (HMC). The HMC model is a unique amalgamation of different ML models, such as K-nearest neighbor (KNN), extra trees (ET), gradient boosting (GB), AdaBoost (AB), linear discriminant analysis (LDA), ), naive Bayes (NB), support vector machine (SVM), random forest (RFlinear regression (LR), and classification and regression tree (CART). The DS2OS dataset is used to evaluate the proposed method's effectiveness. Several performance metrics, including recall, precision, accuracy, specificity, F1 score, detection rate, and ROC are used to evaluate the system's performance. The proposed model successfully distinguishes between normal and attack traffic, achieving an accuracy rate of 99.7% and 99.8%, respectively. To evaluate the effectiveness of the proposed method, its performance metrics were compared to those of other advanced attack detection algorithms. The outcomes demonstrated that the proposed model outperformed other ML and DL-based techniques