Distributed Secure Consensus Control With Event-Triggering for Multiagent Systems Under DoS Attacks

• Published in: IEEE transactions on cybernetics Vol. 51; no. 6; pp. 2916 - 2928
• Main Authors: Yang, Yang, Li, Yanfei, Yue, Dong, Tian, Yu-Chu, Ding, Xiaohua
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.06.2021; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Actuators; Computer crime; Consensus secure control; Control systems; Controllers; Denial of service attacks; denial-of-service (DoS) attack; Denial-of-service attack; event-triggered control; Monitoring; Multi-agent systems; multiagent system (MAS); Multiagent systems; Observers; Schedules; self-triggered control; Telecommunications
• ISSN: 2168-2267; 2168-2275; 2168-2275
• DOI: 10.1109/TCYB.2020.2979342

Consensus control of multiagent systems (MASs) has applications in various domains. As MASs work in networked environments, their security control becomes critically desirable in response to various cyberattacks, such as denial of service (DoS). Efforts have been made in the development of both time- and event-triggered consensus control of MASs. However, there is a lack of precise calculation of control input during the attacking periods. To address this issue, a distributed secure consensus control with event triggering is developed for linear leader-following MASs under DoS attacks. It is designed with a dual-terminal event-triggered mechanism, which schedules information transmission through two triggered functions for each follower: one on the measurement channel (sensor-to-controller) and the other on the control channel (controller-to-actuator). To deal with DoS attacks, the combined states in the triggered functions are replaced by their estimations from an observer. Sufficient conditions are established for the duration and frequency of DoS attacks. To remove continuous monitoring of the measurement errors, a self-triggered secure control scheme is further developed, which combines the system states and other information at past triggered instants. Theoretical analysis shows that the followers in MASs under DoS attacks are able to track the leader and meanwhile the Zeno behavior is excluded. Case studies are conducted to demonstrate the effectiveness of our distributed secure consensus control of MASs.