Energy Scheduling Based Remote Estimation In Cyber-physical Systems

With the rapid progress of information technologies,various types of equipment in the phys-ical world have gradually been interconnected.And thus the cyber-physical systems(CPS)which deeply integrate information space and physical space are birthed.They integrate computing,com-munication and control technologies,and percept,manage and regulate all aspects of human life by optimally allocating resources.CPS have been widely applied in industrial production,smart grid,intelligent transportation,intelligent medical and military,which greatly improved the mode of human production and life.Therefore,they have gained great attention of global researches.Most applications of cyber-physical systems require the remote estimation of system state.For example,the accurate estimation of pollution indexes in environmental monitoring determines the effect of monitoring,and the precise estimation of controlled object's operating state in indus-trial production affects the calculation of control signals and the production efficiency.Therefore,remote estimation problem is one of the basic problems in CPS.Accurate and efficient remote esti-mation is the prerequisite for the comprehensive performance of the system.However,the limited resources of networked nodes,the unreliable communication and security threat bring significant challenge to the research.Recently,the research of remote estimation in CPS based on energy scheduling have been widely concerned and have great progress.Nevertheless,there are still many shortcomings.For example,the current research about the online scheduling of resources using real-time information is lack of consideration of the time-varying communication environment.The perception of attack behavior is mainly limited to the typical and single attack mode.And most of the applications are based on the ideal assumptions for physical objects and communica-tion environments.In view of these shortcomings,this dissertation studies remote estimation from the three aspects including performance optimization,security assurance and technology applica-tion.The main work and contributions are summarized as follows:1.An overview of existing literatures on remote estimation in CPS is provided,and deficiencies of this research field have been analyzed.2.Research on optimal dynamic scheduling strategy of sensor energy for remote estimation.Considering the influence of the time-varying fading channel on the remote estimation per-formance,the problem whose objective is minimizing the expected average estimation error covariance and constraint is the average energy consumption is studied.To solve this prob-lem,a dynamic energy scheduling strategy based on real-time packet loss information is designed by leveraging the Markov decision process theory.And the optimality of the dy-namic schedule is proved theoretically.Furthermore,a sufficient and necessary condition for estimation stability is provided for the case where the sensor can only use low energy at each time to send data.3.Research on stochastic event-based energy scheduling for remote estimation.To accurately portray the priori distribution of remote estimation under event-based scheduling,a stochas-tic event-trigger scheduling mechanism with time-varying parameters is introduced.A more effective online power scheduling is designing by using the innovation and the corresponding minimun mean square estimation algorithm is derived.Additionally,the general parameter sequence optimization problem is proposed,and then an optimal parameter switching strat-egy is given for a class of application with special parameter structure.4.Research on event-based attack against remote estimation and the performance evaluation.Consider an attacker with the diversified ability of compromising the sensor,tampering the scheduling algorithm,eavesdropping the measurement and injecting the false feedback in-formation,an event-based intelligent attack strategy which can be avoiding detection is de-signed.Then from the viewpoint of the attacker,the approximated minimum mean square estimation algorithm is derived and the attack threshold with which the communication rate will not be changed during the attack process is given analytically.In order to evaluate the estimation performance under attack,the stochastic analysis theory is leveraged to prove that the designed attack strategy undermines the estimation stability condition and the asymptotic estimation accuracy.5.Research on energy-efficient target tracking based on remote estimation.In the application scenario where mobile sensors with limited sensing range are used to track a moving target,considering the influence of data loss caused by the unreliable communication,a optimiza-tion problem of minimizing the total sensor's moving distance under the constraint that the sensor can track moving target with a predefined probability is constructed.Then based on the application of the remote estimation technique and the optimal task assignment theory,an energy-efficient cooperative tracking method is proposed,which include three steps:esti-mate and predict target's location,cover target's uncertain area and allocate mobile nodes.In the end,the thesis is concluded and some future research works are discussed.