Performance Analysis And Synthesis For Networked Systems Under Different Kinds Of Data Scheduling Mechanisms

This dissertation is devoted to the performance analysis and synthesis issues for several kinds of classical networked systems under different data transmission/scheduling mechanisms.The considered system models are all the discrete-time cases which mainly include the linear time-invariant systems,the time-varying stochastic parameter systems,the stochastic complex network systems,the multi-agent systems,etc.The data transmission mechanisms of interest consist of the event-triggering scheme,the multi-rate sampling/transmission strategy,the redundant-channel-based transmission mechanism,the coding-decoding-based data transmission approach.Moreover,some new techniques have been developed/put forward to deal with the problems mentioned above,such as the recursive-matrix-inequality-based parameter design method,the dynamic-quantization-based coding scheme,the dynamic differential coding technique,et al.The content of the dissertation is divided into two parts.In the first part,the attention is focused on the event-triggered state estimation problems for several kinds of complex systems(e.g.complex networks,sensor networks and neural networks).By utilizing the Lyapunov stability theory as well as the recursive matrix inequality(RMI)technique,the analysis and synthesis problems for the addressed complex systems are specifically investigated,and several easy-to-implementation estimator design schemes are developed that are effective for the situation of limited network resources.In the second part,the aim of the dissertation is to handle the control and filtering problems for several kinds of network systems via different coding-decoding-based communication strategies.In terms of the dynamic quantization approach,the differential coding method and the Riccati-like equation technique,the mathematical models of the coding-decoding schemes are established.Furthermore,under the proposed coding-decoding schemes,the analysis issues of the system performances(such as the asymptotical/exponential stability,consensus performance,the optimality of the variance of the filtering error)are dealt with,which enrich the applications of the data coding theory in the control community.In addition,the finite-horizon filtering issue is considered for a class of stochastic time-varying multi-rate systems with the redundant-channel-based transmission mechanism and the corresponding filter design scheme is proposed to guarantee that the filtering error system meets the multiple performance requirements.In particular,the main content of this dissertation comprises the following aspects:Based on the event-triggering communication mechanism,the state estimation problem is investigated for a kind of time-invariant complex networks with mixed time-delays.With the aid of the Lyapunov stability theory and stochastic analysis techniques,sufficient conditions are obtained to ensure the exponentially ultimate boundedness of the estimation error in the mean-square.Moreover,the explicit expression of the upper bound of the estimation error is derived and the impacts of the system parameters on such a bound are also analyzed.By using the similar approach,the event-triggered state estimation issue for a class of time-delayed recurrent neural networks is discussed,where a more general triggering scheme is adopted which is dependent on the individual component of the measurement outputs.The estimator design strategy is proposed to guarantee that the estimation error is exponentially bounded.Next,different from the above time-invariant case,the distributed filtering problem is taken into account for time-varying stochastic parameter systems,where the sensor measurements suffer from the successive packet dropouts.Among the sensors,the data interaction are governed by the event-triggering communication strategy,based on which both the disturbance attenuation and the variance-constrained performances of the filtering error system are investigated.For a class of linear time-varying multi-rate systems,the multi-objective finite-horizon filtering problem is considered where the packet dropout is taken into account during the data transmission between the sensor and filter.In order to improve the reliability of the data communication,the redundant-channel-based transmission scheme is employed.With the help of the lifting technique and the RMI approach,the existence condition of the desired finite-horizon filter is obtained and the filter parameters are characterized.Compared with the case of single transmission channel,simulation examples further demonstrate the advantages of the proposed redundant-channel-based filtering algorithm.The observer-based synchronization control problem is studied for a class of discrete-time dynamic networks under a network-resource-constrained environment.Both the dynamical quantization and the differential coding schemes are applied to the observer output of each network node.The randomly occurring packet dropout is also considered during the transmission of the codewords.Based on the received codewords,the decoded value is obtained with certain decoding scheme.By utilizing the decoded state estimation,the synchronization controller is designed for each node By recurring to the input-to-state stability(ISS)theory,a sufficient condition is derived to ensure the asymptotic stability of the synchronization error system.Then,the controller parameters are characterized by solving a certain linear matrix inequality.Furthermore,the observer-based consensus control problem is dealt with for a class of networked multi-agent systems with known directed communication topology.The similar coding scheme is employed so as to reduce the occupancy of the network resource during the data transmission.By means of the graph theory and the ISS theory,sufficient conditions are established to guarantee the asymptotic consensus of the multi-agent systems.By decoupling the augmented multi-agent systems into M individual subsystems,the controller parameters are obtained by solving certain matrix inequality.It is revealed that the derived matrix inequality is dependent on the eigenvalues of the adjacency matrix of the directed communication graph and the number of the matrix inequality is independent of the number of the agents.The recursive filtering problem is considered for a class of discrete-time stochastic nonlinear systems via the multiple description coding(MDC)scheme.The channel fading phenomenon is taken into account during the data transmission between the sensor and the coder,and then the MDC scheme is applied to the faded measurements.The mathematical model of the decoded measurements is established by introducing the Kronecker-delta function and a set of certain stochastic sequences that satisfy specific probability distributions.Based on such a model,corresponding bounds of the decoding error under different cases of the packet dropouts are derived By solving a set of coupled Riccati-like equations,the filter parameters,which guarantee that the upper bound of the variance of the filtering error is optimal at each time step,are obtained.Moreover,the MDC-based control problem is addressed for a class of discrete-time linear systems.Different from the commonly used static coding approach,the dynamic MDC scheme is first proposed,by which the asymptotic convergence of the decoding error is ensured if the encoder/decoder parameters are properly set.Then,the decoded-value-based controller analysis and design issues are addressed and the asymptotic stability of the closed-loop system is guaranteed.In addition,the impacts of the system parameters(e.g.system matrices,coder/decoder parameters and channel parameters)on the stability of the closed-loop system are evaluated.