Distributed H_? Filtering For Nonlinear Systems Based On Sensor Networks

Since the improvement of information processing capabilities,sensor networks(SNs)have been widely applied in physical security,environmental monitoring,patient detection and so on.SNs customarily consist of a rich body of sensor nodes with the ability among sensing and processing information,which enables the system to collect and analyze information momentarily in the monitoring area.Similarly,the research of distributed H_? filtering has been receiving considerable attention in the field of system monitoring as well as signal processing,and the distributed H_? filtering problem based on SNs has become a research hotspot.In view of nonlinear systems,nevertheless,the related research is not adequate enough.In this paper,a variance-constrained filtering scheme is adopted to design distributed H_? filter for nonlinear systems over SNs.The main research contents include the following three parts:(1)The variance-constrained distributed H_? filtering problem is studied for nonlinear systems over SNs subject to multiplicative noise as well as successive packet dropouts.Random variables with zero mean and unit variance are employed to characterize the multiplicative noise,and the successive packet dropouts'phenomenon is described by random variable obeying Bernoulli distribution.Besides,a dynamic event-triggered mechanism is exploited to reduce unnecessary data conflicts as well as fulfill the trade-off between filter performance and communication burden.A distributed filtering error system model is established,and the sufficient conditions are proposed to guarantee that the filtering error system satisfies filtering error variance constraint and H_? performance constraint.Meanwhile,the design method of filter parameters is developed apart from matrix inequalities.The effectiveness of the distributed filter design scheme is lastly confirmed from the perspective of numerical example.(2)For the nonlinear systems over SNs,the issue of the variance-constrained resilient distributed H_? filtering is investigated,where the nonlinearity is stochastic and is characterized by random variable which obeys the Bernoulli distribution.For redundant data,in order to raise the transmission efficiency among sensor nodes,the dynamic event-triggered mechanism with weighting matrix is introduced to preserve limited network resources.Considering that the filter will change under the influences of environment and other factors,the resilient distributed H_? filter is designed,where the gain fluctuation is expressed by additive uncertainty and the randomness of gain fluctuation is characterized by random variables obeying the known probability distribution.Then,the resilient distributed filtering error dynamic system model is derived,and variance constraint and H_? performance constraint of the filtering error are established to ensure that the desired requirements are fulfilled for the filtering error system.What's more,the desired filter gain matrices are resolved through the matrix inequalities,and the feasibility of the designed filter algorithm is verified through simulation results.(3)The variance-constrained distributed H_? filtering problem is discussed for nonlinear systems with randomly occurring deception attacks over SNs.A set of uncorrelated random variables subject to Bernoulli distribution are used to describe deception attacks in data transmission over SNs.The impacts of deception attacks and static event-triggered mechanism are further analyzed on the filtering performance.By using stability theory and variance constraint idea,variance constraint and H_? performance constraint of the filtering error are derived which are satisfied by the filtering error system,and the explicit expressions of distributed H_? filter gain matrices are given.In the end,it can be shown by the results of simulation that the static event-triggered filter method which proposed is feasible.