Distributed Filtering With Incomplete Measurements Over Wireless Sensor Networks

Wireless sensor network is a multiple-hop network consisting of a large number of sensor nodes,sensors usually communicate by way of self-organization.Since wireless sensor network technology is widely used in various fields,filtering problem as one of the basic problem has important research significance.One of the most important characteristics in wireless sensor networks is that each sensor is equipped with limited computing and power resources.The major power consumption in sensor nodes comes from the communication process.It is necessary to save the energy by reducing the data transmission in designing distributed filter algorithm.Meanwhile,the usage of the network gives rise to missing measurements.How to deal with it is the key to the distributed filter issue.The main contents of this thesis consist of the following two parts:First,the distributed H_?filtering problem is studied for a class of linear systems with multiple missing measurement.A set of stochastic variables are introduced to model the probabilistic missing measurements.In the system,all of the stochastic variables are mutually independent but satisfy the Bernoulli binary distribution.Sufficient conditions for the existence of the distributed H_?filter are derived in terms of linear matrix inequalities,which ensures the asymptotic stability as well as a prescribed H_? performance for the filter errors.A simulation example is exploited to demonstrate the effectiveness of the proposed design procedures.Second,The problem of distributed H_? filtering for a class of discrete systems with random packet drop under an event-triggered communication mechanism is studied.Event-triggered communication mechanism is introduced to reduce burden of communication and energy consumption of wireless sensor network,the sensor broadcasts its measurements to the neighboring nodes only when the certain trigger condition is satisfied.A filtering error system model is established to account for both the event-triggering condition and data dropout in a unified framework.A performance criterion for distributed H_?filtering and a design method of filtering parameters under the event-triggered communication scheme are given in terms of linear matrix inequalities.Finally,a numerical example is given to illustrate the effectiveness of the proposed design scheme.