Every day, a considerable number of new cybersecurity attacks are reported, and the traditional methods of defense struggle to keep up with them. In the current context of the digital era, where industrial environments handle large data volumes, new cybersecurity solutions are required, and intrusion detection systems (IDSs) based on artificial intelligence (AI) algorithms are coming up with an answer to this critical issue. This paper presents an approach for implementing a generic model of a network-based intrusion detection system for Industry 4.0 by integrating the computational advantages of the Nvidia Morpheus open-source AI framework. The solution is modularly built with two pipelines for data analysis. The pipelines use a pre-trained XGBoost (eXtreme Gradient Boosting) model that achieved an accuracy score of up to 90%. The proposed IDS has a fast rate of analysis, managing more than 500,000 inputs in almost 10 s, due to the application of the federated learning methodology. The classification performance of the model was improved by integrating a generative adversarial network (GAN) that generates polymorphic network traffic packets.

Moving to the cloud is a topic that tends to be present in all enterprises that have digitalized their activities. This includes the need to work with software environments specific to various business domains, accessed as services supported by various cloud providers. Besides provisioning, other important issues to be considered for cloud services are complexity and performance. This paper evaluates the processes to be followed for the deployment of such a software environment in the cloud and compares the manual and automated methods in terms of complexity. We consider several metrics that address multiple concerns: the multitude of independent paths, the capability to distinguish small changes in the process structure, plus the complexity of the human tasks, for which specific metrics are proposed. We thus show that the manual deployment process is from two to seven times more complex than the automatic one, depending on the metrics applied. This proves the importance of automation for making such a service more accessible to enterprises, regardless of their level of technical know-how in cloud computing. In addition, the performance is tested for an example of an environment and the possibilities to extend to multicloud are discussed.

This paper investigates the integration of intrusion detection systems (IDSs) within Digital Twin (DT) architectures to enhance cybersecurity in industrial environments. Using the CICIDS2017, CIC Modbus, and 4SICS 2015 datasets, we evaluate the performance of Random Forest (RF) and Support Vector Machine (SVM) in detecting network intrusions. Results indicate that RF achieves an accuracy of 99.9% for CICIDS2017, with high precision, recall, and low false positives. In contrast, SVM exhibits an accuracy of 94.2% for the same dataset, struggling with high rates of false positives and moderate recall. Similarly, for 4SICS 2015, RF demonstrates an accuracy of 93%, being balanced and reliable for industrial applications, while SVM shows only 88% accuracy, with a low precision of 65% and a high false alarm rate. For the CIC Modbus dataset, RF displays an accuracy of 95% in validation and 93% in testing, highlighting strong detection in ICS networks. However, SVM maintains an accuracy of 88%, with weak separation between benign and malicious traffic, and a higher misclassification rate. Our findings highlight the importance of DT-IDS integration in real-time threat detection and system resilience, paving the way for future research in deep learning-based IDS solutions.