Problem Statement

An important recent trend is that the connection between the Internet and elements of the physical world (such as machines, robots, cars, energy facilities) is getting stronger and stronger. The resulting cyber-physical systems enables dramatic perspectives such as the Industry 4.0 or energy smart grids.

The energy transition leads towards smarter electric power systems taking the form of cyberphysical systems in which the electrical power grids are strongly interlinked with a growing number of information and communications systems. Recent attacks on oil pipelines, electricity production sites, etc. have highlighted the crucial importance of securing these critical infrastructures and to develop system-wide security strategies.

Contribution: machine learning-based detection and adversarial attacks

Recent studies have shown that Machine learning in general and deep learning specifically are vulnerable to adversarial attacks where the attacker attempts to fool models by supplying deceptive input. We investigate the transferability of adversarial network traffic against multiple machine learning-based intrusion detection systems.

In this research, we developed a defensive mechanism to limit the effect of the transferability property of adversarial network traffic against machine learning-based intrusion detection systems in the context of the Internet of Things.

Contribution: the energy smart grids

This research work aims to develop new knowledge, methods and tools necessary to guarantee the cybersecurity of electrical supply via the transport network, while taking into account the specific nature of cyber threats induced by the communication subsystems. Solutions focus on preventive (security and detection) and curative (recovery and resilience) measures. Preventive measures will include the use of ad-hoc architectures and security protocols for the critical systems of production sites, Transmission System Operators (TSOs), Distribution System Operators (DSOs) and market operators. As a part of the research a  proof-of-concept implementation is developed with and industrial partner.