Research On Distributed Estimation Algorithm For Networked Stochastic Uncertain System With Network Attack

With the development of computer and communication technology,distributed control system,as one of the basic research frameworks of networked control system,has the characteristics of low energy consumption,high robustness and high estimation accuracy.At the same time,the distributed structure leads to insufficient protection during the communication transmission between nodes,which is more vulnerable to malicious network attacks.Therefore,the problem of secure distributed estimation under network attacks has become a hot topic.This paper mainly studies the distributed estimation of networked random uncertain systems with deception attacks,denial of service(DoS)attacks,and both deception attacks and DoS attacks.The main research contents are as follows:For networked linear discrete stochastic uncertain systems with multiplicative noise and deception attacks,a distributed optimal estimation algorithm in the sense of linear unbiased minimum variance and a maximum-minimum robust estimation algorithm are presented.Considering the fading of attack signals caused by engineering physical constraints,an attack model with random attenuation is given.Based on the given attack model,firstly bring the measured data of sensor nodes themselves and those of neighboring nodes that may be attacked into the distributed filter algorithm,and then a distributed optimal filter is obtained based on the linear unbiased minimum variance estimation criterion.When the true variance of attack noise is unknown but the upper bound is known,a robust distributed filter with the upper bound of the variance of the conservative attack noise is given based on the minimax robust estimation method,and the robustness of the robust distributed filter is proved by using the dynamic error system analysis method.For networked linear discrete stochastic uncertain systems with multiplicative noise and DoS attack,three distributed estimation algorithms are presented.First,a set of Bernoulli random variables is used to describe the blocking of sensor transmission channels caused by DoS attacks.Then,based on the linear unbiased minimum variance estimation criterion,the distributed filtering algorithms depending on the time stamp of attack occurrence or the probability of attack occurrence are given respectively.Finally,the stability of the algorithm is proved by applying the dynamic error system analysis method and the dynamic variance error system analysis method.For networked linear discrete stochastic uncertain systems with deception attacks,DoS attacks and multiplicative noise,a distributed optimal estimation algorithm in the sense of linear unbiased minimum variance and a maximum-minimum robust estimation algorithm are presented.Based on the linear unbiased minimum variance estimation criterion,a distributed optimal filter is presented when the system when the system is attacked by both deception attacks and DoS attacks.By applying the dynamic error system analysis method and dynamic variance error system analysis method,it is proved that the distributed optimal filter converges to the corresponding steady-state distributed optimal filter;When the true variance of attack noise is unknown but the upper bound is known,the robust distributed filter algorithm is studied based on the minimax robust estimation method,and the robustness of the robust distributed filter is proved by using the dynamic error system analysis method.