Research On Networked Multi-agent Security Consensus Control Strategy Based On Set Membership Estimation Under Spoofing Attack

In recent years,with the development of communication technology and the complexity of industrial scale,the consensus and security control of networked agents has gradually attracted attention.However,interference in the network transmission environment and network attacks seriously threaten the consensus and security of networked agent systems,especially more stealthy and destructive deception attacks.Thus,how to find an effective estimation method and estimate the state of the system under deception attacks is crucial to the research on the consensus and security of networked agents.Traditional methods such as Kalman estimation require prior knowledge of the mean and variance of the attack signals in advance.Set-membership estimation only assumes that the deception attack is an unknown but bounded(UBB)signal,which avoids the assumption of prior knowledge of attack signals and is easy to describe.Therefore,in this paper,the tracking control of single-agent systems based on set-membership estimation and the consensus and security control of leader-follower multi-agent systems based on set-membership under deception attack are realized.In this paper,we first study the tracking control problem of single-agent systems under deception attacks.The deception attack suffered in the data transmission channel is modeled as an unknown but bounded signal(UBB),and a tracking control framework based on a novel set-membership estimation is proposed.Both the real state and any estimated state of the system exist in the ellipsoid estimation set.Then the tracking control problem is further transformed into the stability analysis of the error closedloop system.Through the Lyapunov function,the sufficient conditions to make it stable and the solution algorithm of the estimator and controller based on linear matrix inequalities are obtained.Finally,a convex optimization method is used to obtain the best set of estimates.Numerical simulations demonstrate that the proposed method can achieve effective tracking of single-agent systems.For the tracking control of the leader agent by multiple follower agents,it can be further regarded as the consensus and security problem of multi-agents.This paper considers the time-varying topology due to the existence of position movement and communication disconnection.The deception attack suffered between each follower sensor-to-controller and controller-to-actuator is modeled as an unknown but bounded signal.A distributed set-membership estimation control strategy is proposed,in which the controller of a single subsystem receives arbitrary estimates of itself and neighboring estimators.The ellipsoid estimation set of each follower contains both the arbitrary estimated value and the true value of the system.The sufficient conditions for the existence of the estimator and the controller are obtained through the Lyapunov function and the stochastic control method,and a convex optimization algorithm is established to find the optimal ellipsoid estimation set.Numerical simulations prove that the control algorithm in this paper can effectively achieve the consensus and security of multi-agent systems under deception attacks.To sum up,the novel set-membership estimation method proposed in this paper only needs to know the upper and lower bounds of the attack,and the estimation results use the estimated set instead of the estimated points.It has been applied in the consensus and security problem of single-agent system and multi-agent system under deception attack,and effectively guarantees the consensus and security of the system.