Robust Attack Detection Approach for IIoT Using Ensemble Classifier

Generally, the risks associated with malicious threats are increasing for Internet of Things (IoT) and its related applications due to dependency on minimal resource availability IoT devices. Thus, anomaly-based intrusion detection models networks vital. Distinct methodologies need be developed Industrial (IIoT) network as threat is a significant expectation stakeholders. Machine learning approaches considered evolving techniques that learn experience, such have resulted in superior performance various applications, pattern recognition, outlier analysis, speech recognition. Traditional tools not adequate secure IIoT use protocols industrial systems restricted possibilities upgradation. In this paper, objective develop two-phase anomaly model enhance reliability an network. first phase, SVM Naïve Bayes, integrated using ensemble blending technique. K-fold cross-validation performed while training data different testing ratios obtain optimized test sets. Ensemble uses random forest technique predict class labels. An Artificial Neural Network (ANN) classifier Adam optimizer achieve better accuracy also used prediction. second both ANN results fed model’s classification unit, highest value final result. The proposed tested standard attack datasets, WUSTL_IIOT-2018, N_BaIoT, Bot_IoT. obtained 99%. A comparative analysis state-of-the-art demonstrate superiority results. outperforms traditional thus improves