Risk Assessment of Multi-Agent System Under Denial-of-Service Cyberattacks Using Reachable Set Synthesis
Primary Investigator:
Inseok Hwang
Minhyun Cho, Sounghwan Hwang
Abstract
Multi-agent systems (MASs) have vulnerabilities to various types of cyberattacks disrupting inter-agent communication. To assess the potential risk associated with these cyberattacks, this paper proposes a proactive risk assessment method using reachable set synthesis. Denial-of-Service (DoS) attacks, where adversaries can disrupt communication by a sequence of link disconnections (dynamic alterations), are specifically considered. Our method employs the calculation of reachable sets using Lyapunov functions and linear matrix inequalities (LMIs) derived from them. The proposed method can evaluate the risk of DoS attacks for both individual agents and the entire system in two levels by computing over-approximated ellipsoidal reachable sets. To demonstrate the applicability of our method, we provide an illustrative example involving a leader-follower MAS performing formation control in an adversarial environment with scattered obstacles.