Distributed Estimation Of Complex Systems Over Sensor Networks

Sensor networks are widely applied in many fields owing to the low cost and customizable of sensors.However it is found that in practical engineering applications,some random occur-ring phenomena are inevitable in sensor networks due to the influence of serve environment and aging or failure of sensor devices,such as random gain variations and random nonlinearities,which can cause the oscillations and instability of the systems.Thus,it is vital to consider these factors when investigating distributed estimation/filtering.In addition,it should be noted that some networked phenomena related with sensor networks,such as network-induced delays and missing measurements,are non-ignorable.Therefore,it is significant to discuss distributed estimation of complex systems over sensor networks.This thesis focuses on the problem of distributed estimation of complex systems over sen-sor networks,including non-fragile H? filtering and l2-l? filtering with random gain vari-ations,and exponential estimation for nonlinear systems through Round-Robin protocol.The main contents of this thesis can be summarized as follows:First,the problem of distributed non-fragile H? filtering of discrete-time systems with randomly occurring gain variations is considered.The additive random variations of filter gain are investigated.Based on Lyapunov function method,the H? performance of the augmented filtering error system is analyzed.Then,a distributed non-fragile H? filter is designed and the sparsity of the weighted adjacency matrix is solved.Second,the problem of distributed non-fragile l2-l? filtering over sensor networks with random gain variations and fading measurements is discussed.A random sequence with any known statistical information is applied to describe the probabilistic properties of the fading measurements.B ased on Lyapunov function method,the l2-l? performance of the augmented filtering error system is investigated,and the design method of distributed non-fragile l2-l?filter is proposed.Third,the distributed exponential estimation problem for nonlinear systems with Round-Robin protocol is investigated.In order to solve the problem of network congestion in sensor networks,Round-Robin protocol is introduced in this part.One-side Lipschitz nonlinearity,which is more general than standard Lipschitz condition,is considered.By constructing an ex-ponential Lyapunov function,the sufficient condition is established to guarantee the nonlinear augmented system to be mean-square exponentially ultimately bounded.And the exponential decay coefficient,decay rate and the upper bound can be obtained.Then,a distributed expo-nential estimator is designed.Fourth,the summary and further directions are given.