|
This project develops a novel methodology for designing secure cyber and cyber-physical systems that can detect attackers and protect against malicious behavior after the system has been compromised.
Networked control systems contain actuators and sensors that can be compromised by a malicious attacker intent on altering the system behavior by corrupting sensor values or actuator commands, thereby enabling or forcing the execution of unsafe behavior. The PIs investigate the intrusions that resulted in compromised system components (sensors or actuators) and design remedial control strategies that not only combat malicious attackers upon detection of intrusion but are also resilient to potential intrusions, thereby preventing damage from being inflicted upon the system.
The technical approach has its foundations in the theories of diagnosability, opacity, and supervisory control of discrete-state event-driven dynamic systems. The research plan to the problems of intrusion detection and resilience against attacks contains a creative blend of these theories with game-theoretic approaches to reactive synthesis problems, in order to capture the dynamic game between the control system (acting as the defender) and the attacker. The goal of the defender is to optimize its trade-off between functionality and vulnerability. Given the prevalence of feedback control loops in cyber and cyber-physical systems, the methodology being developed will impact a large class of technological systems that are of great societal importance.
|
TWC: Small: Intrusion Detection and Resilience Against Attacks in Cyber and Cyber-Physical Control Systems