H-infinity Filtering Of Network With Time-delay Or Packet Dropout And Research On Mobile Multi-sensor Management Approachs

With the progress of science and technology, sensor networks are widely used inmilitary and civilian fields. Along with the development of communicationstechnology, computer technology and control theory, research and application ofsensor networks are receiving more attentions.In data transmission, sensor networks often arise time-delay and packet lossproblems, therefore, this paper proposed H-infinity filter design for discrete-timesystem with random time-delay and packet dropout. Besides, the management ofsensor network still exist a series of key issues in tracking, detecting and other aspectsof the applications. We proposed the selection and planning schemes of mobilesensors for the research of multi-sensor application in target tracking.This thesis carrys out the following research work.First, a method of H-infinity filter design for a class of discrete-time systems isproposed for the stochastic time delay and packet dropout in the network. Time-delayand packet dropout are assumed to be two Bernoulli stochastic sequences. Sufficientconditions are derived in terms of linear matrix inequalities (LMI) for the existence ofthe filter. The system of filtering errors for the designed filter is exponentiallymean-square stable and satisfy a prescribed H-infinity performance.Second, a scheme of selecting and configuring algorithm based on the linearprogramming for the mobile multi-sensor with state-constrained is given inmulti-target tracking. We set three controlling input-modalities for each sensor. Wethus can select approximate optimal sensors and their input modalities, which solvethe interaction between sensor selection, sensor modalities and the tracking processing,so as to acquire effective tracking of multi-target.Third, a strategy of sensor selection and planning based on linear programming isput forward for the selection and direction planning of mobile multi-sensor in singletarget tracking. The strategy first adopted discretization strategy and optimizingstrategy to derive the moving directions of the sensors, respectively. The linearprogramming is further imported to select and planning mobile multi-sensor, throughwhich we can achieve effective tracking of target.