ENHANCING MODELLING OF CYBER-SECURITY FRAMEWORK FOR CRITICAL INDUSTRIAL INFRASTRUCTURE USING MACHINE LEARNING TECHNIQUE

Subject Area: CONTROL SYSTEM ENGINEERING

Over the years the critical industrial infrastructure has suffered many problems of which cyber-attack is among the major concerns. This attack exploits hidden vulnerabilities on the network and penetrates with threat for ransomware. To address this problem, this study presents enhancing the modelling of cyber security framework for critical industrial infrastructure using machine learning technique. The aim is to model a proactive defence mechanism capable of identifying attacks effectively in Industrial Internet of Things (IIoT). The research method used are vulnerability assessment and identifying potential exploits or attack, data collection of cyber threats dataset, data process through normalization and feature extraction, multi-layered neural network, back-propagation algorithm, Intelligent Cyber Threat Detection (ICDS), and cyber threat detection response system. The research design approach used mathematical and structural methods to model the ICDS and the integration of the response system on the IIoT network and then tested the model using MATLAB programming software. The ICDS was evaluated considering key parameters for cyber security success such as precision, recall, accuracy, sensitivity, specificity and latency. The results for precision reported 96.9%, recall was 96.7%, accuracy was 96.9%, sensitivity reported 98.2%, specificity 94.3% and latency: 70.82%. These results implied that the ICDS was able to detect threats penetration on the network and mitigate it with high success rate. To validate the ICDS model, comparative approach was applied considering the IIoT network without IDCS and the improved network with integrated ICDS. The result reported 75.65% improvement for loss, 21.80% improvement for throughput, bandwidth utilization factor reported 46.46% improvement and finally latency reported 70.82% improvement. In conclusion the study identified potential vulnerabilities in IIoT and made recommendations to patch them, while developing an ICDS for the detection of cyber threat and minimizing the impact on IIoT.