Resilience Against Sensor Deception Attacks at the Supervisory Control Layer of Cyber-Physical Systems: A discrete event systems approach

Abstract:

Although Cyber-Physical Systems (CPS) are already ubiquitous in our society, their security aspects were only recently incorporated into their design process, mainly in response to catastrophic incidents caused by cyber-attacks on CPS. There is now a critical need to provide techniques for analyzing the behavior of CPS while under attack and to synthesize attack-resilient CPS. In this dissertation, we address CPS under the influence of an important class of attacks called sensor deception attacks, in which an attacker hijacks sensor readings to inflict damage to CPS. The formalism of regular languages and their finite-state automata representations is used to capture the dynamics of CPS and their attackers, thereby allowing us to leverage the theory of supervisory control of discrete event systems to pose our investigations.

We focus on two questions:

• Can we automatically find sensor deception attacks that damage a given CPS?

• Can we design a secure-by-construction CPS against sensor deception attacks?

Using techniques from the fields of supervisory control, graph games, and Markov decision processes, this dissertation provides a comprehensive set of algorithmic techniques to analyze and mitigate sensor deception attacks at the supervisory layer of CPS.

Chair: Professor Stéphane Lafortune

REMOTE: https://umich.zoom.us/meeting/register/tJcscOyqr
DgtE9D8_2324Lzoc-xVLG2NMFdK