Mitigation of OT/IT Cyber Attacks Using Machine Learning Modelling Techniques

Abstract

Cybersecurity is the need of the hour as due to rising Cyber threats OT (Operational Technology) / IT infrastructure is being compromised and there is revenue loss in order of million dollars. Various power sectors are being hit by malicious software and their data and sensitive information is being compromised. Goal of Cybersecurity is to achieve Confidentiality, Integrity, and availability. Due to Rising Ransomware, Trojan Horse and spyware malware OT infrastructure is being effected. For implementation of machine learning model, a PI system implementation was carried out to bring the real time network diagnostic signal values to set up program consisting of Machine Learning algorithms installed in lab setup PC.

Since the developed model outcomes were impeded by the limitation of sufficiency of data set for the model training and testing during the initial stages of PI system data collection of diagnostic signals, the developed algorithms have been tested on similar data sets available on on-line technical sites. The results of the preventive failure predictive algorithms are very much closure to actual failures for the systems under study.

Together with some practical cases of applications, the idea of design of implementation strategy and importance of commercial industrial assets are introduced in the paper. Characteristics and prospects of application of Machine Learning has been analyzed. The outcomes of the study are produced for proactive maintenance in the form of replacement/ bug-fixing in the OT Network much before any catastrophic failure.

Use of Machine Learning concepts in the predictive maintenance or for determining the residual life as demonstrated in the paper goes a long way in safe and optimized operation of commercial industrial assets. Model developed can be part of all upcoming distributed control system (DCS) designs.