Optimal recovery from attacks against cyber-physical systems

The rising frequency of cyber attacks necessitates automated attack recovery.

This project will address this need by framing attack recovery for cyber-physical systems as a dynamic game, i.e., a strategic interaction between the attacker and the system operator. To solve this game, we will develop a novel method that combines causal inference with game-theoretic optimization.

Our method is based on the insight that effectively mitigating cyber attacks requires a cause-and-effect understanding of recovery measures.

Current methods for attack recovery lack this understanding and ignore the causal structure, which makes them inefficient. The experimental part of this project will include a testbed where real attacks and recovery actions can be executed.

Unlike the simulation studies from previous research, this experimental setup will offer practical insights.The first 18 months of the project will be carried out at the University of Melbourne with Prof. Alpcan, a leading authority on game-theoretic approaches to security. This phase will allow me to expand my competencies in formal methods and game theory.

These areas complement my PhD, which focused on security engineering.

Insights gained during this period will lay the groundwork for the subsequent 18 months, which will be carried out at Imperial College London with Prof. Lupu, an expert in causal inference and policy-based security management. During this phase, the focus will shift to the implementation of the method.