Distributed Fusion Estimation For Multi-Rate Multi-Sensor Networked Systems Under Cyber Attacks

Multi-sensor networked fusion estimation systems have attracted wide attention in many fields because of their outstanding information acquisition and collaborative processing capabilities.Compared to centralized estimation structures,distributed fusion estimation has made some research progress due to its advantages of parallel processing capabilities and reducing the computational burden.However,the dynamic characteristics of different physical objects are different and the complexity of their surrounding environment,so it is difficult for different sensors to maintain the same sampling rate.On the other hand,the information transmission of multi-sensor networked fusion estimation systems is affected by both open network and communication bandwidth,so the multisensor networked systems are vulnerable to cyber attacks,packet loss,and time delays.However,there are few studies on distributed fusion estimation based on multi-rate multisensor networked systems simultaneously considering the impact of cyber attacks and communication bandwidth constraints.Considering that feedback from the fusion center can reduce the impact of attacks,then the event-triggered mechanisms can be designed to reduce the communication burden and avoid unnecessary transmission of feedback.This paper mainly studies the distributed fusion estimation of multi-rate multi-sensor networked systems under cyber attacks.The specific contents are as follows:1)For a class of multi-rate multi-sensor networked systems with multi-channel timevarying observation delays subject to hybrid attacks,a novel multi-rate resilient distributed fusion method with delayed output is proposed.The local optimal Kalman filters based on the multi-step smoothers are designed respectively for the sub-systems with different observation delays in the linear minimum variance sense by the innovation analysis approach.A local estimation compensation strategy based on the multi-modal characteristics of hybrid attacks is constructed when there are both denial of service(DoS)attacks and deception attacks launched from the estimator to the fusion center.Considering the asynchrony of local estimation,a fusion matching mechanism for local estimation is designed,and a distributed fusion estimation method based on covariance cross fusion(CI)and matching strategy is further proposed.Moreover,the designed fusion method can improve the estimation accuracy and reduce the computational burden.Finally,the effectiveness of the proposed algorithm is verified by numerical simulation.2)The distributed fusion estimation problem for multi-rate multi-sensor networked nonlinear systems with multi-channel DoS attacks is studied,where DoS attacks are launched during the information transmission process from local sensors to local estimators.The local Kalman estimators with fusion center feedback are designed respectively for different subsystems based on unscented Kalman filtering(UKF)by considering the impact of DoS attacks on system estimation performance.In order to reduce the communication burden,an event-triggered transmission strategy is introduced into the feedback mechanism,which can effectively reduce the feedback of redundant information.Then a multi-modal distributed fusion estimation method based on local estimation with event-triggered feedback is designed in the fusion center.The compensation strategy and matching mechanism based on the one-step prediction of fusion estimation are proposed to aim at the asynchrony of local estimations.A multi-modal distributed fusion method based on different characteristics of compensated local estimation is further designed,in which CI fusion and weighted matrix fusion are used for switching.The designed fusion method fully considers the local estimation characteristics after compensation and can improve the estimation accuracy.Finally,a simulation example verifies the effectiveness of the proposed method.