Abstract

The Industrial Internet of Things (IIoT) acceleration has caused automation and data exchange enhancements within industrial surroundings. Yet, this evolution has presented security issues due to the raised exposure of critical infrastructure to cyber threats. This study focuses on designing a thorough model for identifying and mitigating vulnerabilities within IIoT networks utilizing Machine Learning (ML) and Deep Learning (DL) techniques. A replicated IIoT network infrastructure was set to communicate and exchange data for simulation. Use of Python script to execute network scanning and data collection, distinct possible vulnerabilities. Then, ML-DL analysis is handled by employing techniques of gradient boosting, logistic regression, decision trees, random forest, multilayer perceptron, and convolutional neural network. Throughout, gradient boosting has proven to higher performance accuracy rate in recognizing the most impactful vulnerabilities. As well as a model integral part of a Cost-Benefit Analysis (CBA) provides security recommendations to mitigate identified vulnerabilities. According to the CBA model vulnerabilities are prioritized based on the severity, related costs, and potential benefits of mitigation. The proposed Cybersecurity model in addition to high accuracy in vulnerability detection also provides a standardized approach for categorizing Cybersecurity countermeasures according to cost-effectiveness. This study emphasizes the need for a consolidated Cybersecurity model for the IIoT and shows the capability of ML techniques to advance Cybersecurity posture. Future work considered testing the model in a real operation environment of IIoT, refining the model, and enrichment with more knowledge base actionable mitigations.

Statistics