Event-triggered Consistency Control Of Multi-agent Systems Under Malicious Network Attack

With the development of the network control technology and the embedded technology,the concept of multi-agent systems(MASs)emerged as the times require.MASs have received more and more attention from scholars because it can accomplish complex group tasks that cannot be completed by a single agent.The consensus problem is the most widely explored topic in the field of MASs.Designing a distributed control protocol based on network topology is an important task of the consensus problem,which has a direct impact on the clustering,formation and tracking effects of MASs.However,the vulnerability of open networks and the dependence of MASs on communication networks jointly determine that MASs are vulnerable to malicious network attacks,which can seriously affect the consensus effect of MAS.Therefore,it is necessary to consider the impact of network attacks when designing distributed control protocols.In addition,the existing results of consensus theory are mostly focused on continuous-time or fixed discrete-time systems,and too much continuous communication will cause the increase of network resource occupancy and sensor energy consumption.Therefore,this thesis introduces an event-triggered mechanism to design distributed control protocols with the aim of reducing resource waste while achieving consensus.Under the above research background,this thesis designs control schemes based on event-triggered strategies for the consensus problem of MASs affected by different types of malicious network attacks.The main research contents of this thesis are as follows:(1)For denial-of-service(DoS)attacks,this thesis studies the event-triggered consensus control problem of MASs in directed network topology.Considering the occasional Do S attacks launched by malicious attackers,the communication channel may be interrupted.An observer-based event-triggered safety control scheme is proposed to determine the controller update time,and a less conservative separation method is used to design the controller and the observer.Then the upper bounds on the frequency and duration of Do S attacks that can be tolerated by the MASs using the designed control scheme are analyzed under the premise of achieving the consensus objective.Furthermore,with the help of the proposed event-triggering conditions,a strictly positive minimum inter-event time is designed for each agent to eliminate Zeno behavior.Finally,the correctness of the theoretical analysis is verified by numerical simulation.(2)For the false data injection(FDI)attacks with known upper bound of attack energy,this thesis studies the event-triggered secure synchronization control problem of MASs in directed network topology.In the process of information transmission,this thesis assumes that both the control signal and the measurement signal may be tampered with false data by the attacker,and introduces Bernoulli random variables to describe FDI attacks that occur randomly on the measurement output signal and the control input signal.In this thesis,an event-triggered mechanism that depends on the observer output threshold is designed to reduce the waste of network resources.On this basis,an observer-based resilient control scheme is proposed to solve the secure mean-square bounded synchronization problem of directed network topology MASs caused by FDI attacks,and the linear matrix inequality technique is used to design the controller gains and observer gains.Furthermore,the Zeno behavior is effectively eliminated by the use of the inverse proof method.Finally,the validity and applicability of the theoretical results are verified by comparative simulation experiments.(3)For FDI attacks with unknown attack energy,this thesis studies the event-triggered secure synchronization control problem of MASs.Assuming that the attack energy is unknown,the attacker affects the performance of the observer by tampering with the output signal sent by the sensor,thereby further destroying the synchronization effect of the multiagent system.In order to deal with more generalized attackers,this thesis introduces an event-triggered mechanism to reduce the transmission of output signals,and designs a compensation mechanism for attack signals with unknown energy upper bound,so as to reduce the impact of the observer.A resilient control scheme based on adaptive compensation protocol is proposed to solve the secure mean-square bounded synchronization problem of MASs caused by FDI attacks,and the sufficient conditions for realizing synchronization control are given by the linear matrix inequality technique.Finally,the effectiveness and applicability of the resilient control scheme of the designed adaptive compensation protocol are verified by simulation experiments.