Research On Multi-sensor Multi-target Allocation Algorithms Based On Time-varying Measurement Variance

Sensor management in the radar network system is mainly used to improve theoverall performance of the system by optimally allocating the limited senor resource,which can effectively deal with the problems in the complex battlefield environment.However the sensor resources allocation problems become more complex as thenumber of targets increase and the sensor resources and communication resources arelimited. So the thesis has studied the multi-sensor allocation problem for trackingmulti-targets to improve the overall tracking quality. The main contributions aresummarized as follows:1. A tracking-feedback framework based sensor source management isconstructed based on analyzing the traditional data fusion methods and sensormanagement algorithms. Various allocation methods are developed under thisframework.2. A new method of multi-sensor multi-target allocation problem based oncovariance control under time-varying measurement variance is presented toovercome the shortcoming of the estimating covariance in the traditional sensormanagement based on the covariance control algorithm in which the measurementvariances are assumed fixed. The optimal allocation model based on the covariancecontrol is first given by analyzing the tracking quality and the measurement variance.A heuristic algorithm is then proposed to solve the problem quickly. At last, thesimulation result has been analyzed.3. A multi-sensor multi-target allocation problem with time-varyingmeasurement variance based on the information gain is developed from theinformation theory to overcome the shortcoming of the fixed measure varianceassumption and linear filter algorithm in traditional sensor management. This methodhas used the nonlinear filter to estimate the target state and the allocation model isestablished with considering the tracking capacity limitation and detecting distancelimitation. An integer programming method and an improved Hungarian algorithm arethen presented to obtain the optimal sensor resource allocation to meet differenttracking requirements. At last, the simulation has been proposed to prove theeffectiveness of the two algorithms. 4. A multi-sensor multi-target allocation problem with considering thecommunication limitation is presented, in which the communication capacity isconsidered. A quick method based on the expected information gain is then given.This method uses fewest sensors to keep the desired tracking quality. A simulation hasbeen proposed to prove the effectiveness of the method.