Research On Security Estimation Of Cyber Physics Systems Based On Game Theory

With the progress of society,digitization,networking and intelligence have become the mainstream of era,which affects life,production and management of human society.With the improvement of information technology,technologies on computing,communication and control are constantly integrated to realize interconnection of ternary space for physical space,information space and human society.Then Cyber Physical Systems(CPS)is emerged with deep integration ternary space.In view of existence of networks,CPS often presents characteristics of openness,but also brings many potential security threats.Malicious network attacks increase difficulties in designing security estimation algorithm,and even make reliable operations of CPS unguaranteed.Therefore,network attack modeling,state estimation algorithm and defense strategy design are studied in this paper aiming at above problems in CPS.The specific research contents are summarized as follows:Firstly,problems on network attack modeling and system defense strategy design are designed when CPS is subjected to Denial-of-Service(DoS)Attack.Considering that both attacker and defender are energy-constrained,an interaction decision between defender and attacker is described based on game theory.A Nash equilibrium solution is calculated to further obtain a delivery rate of channel by a matrix game method.Then optimal defense strategies and attack strategies are obtained respectively for defenders and attacker by extending the research method to CPS composed of multiple-subsystem.Secondly,problem on security estimation is studied in a state-coupling CPS.Both state estimations and coupling state estimations for system are obtained using an unbiased minimum variance estimation algorithm.At the same time,communication channels are vulnerable to malicious smart attacks which can adaptively adjust transmission power.Pay-off functions are established for smart sensor and smart jamming,smart sensor and smart jammer and a Stackelberg game framework is established to describe interaction decision between smart jammer and smart sensor.The best response strategy of smart jammer and optimal strategy of smart sensor are got and then the existence and uniqueness are given for Stackelberg equilibrium solution.Finally,a problem on security estimation is studied for CPS in the presence of unknown input and smart jamming attack.Using measurement information detected,local optimal state estimations are obtained for smart sensor based on unbiased minimum variance estimation theory.Meanwhile,to save communication cost,a remote state estimator based on event triggering mechanism is designed.An interaction decision between smart attackers and smart sensors is established by Stackelberg game theory to obtain the best response strategy and optimal strategy for smart attacker and smart sensor,respectively.Moreover,sufficient conditions are given for the stability of the system.