Weighted statistical model selection for multi model particle filters

The algorithm speed is the most important fact for tracking of radar targets. Because it requires a real time follow-up for targets motion. In the most preferred algorithm of multi model particle filter (MMPF) for target tracking, the number of calculations for the number of particles and the maneuvering model selection is the most important parameter for determining the process speed of filter. The particle number and/or model calculations should be reduced as much as possible so the reduction of these described two facts expedites the algorithm and eases a real time follow-up. In this study, a new approach which called weighted statistical model selection (WSMS) algorithm is proposed for reduction of model calculations and the results are presented about the applications preformed on MMPF. For evaluate the success of proposed algorithm, in simulations preformed on different scenarios, three different MMPF are used. One of them is a new MMPF which is proposed by us for this thesis. The WSMS is integrated into two of these filters in simulations and the obtained results are compared based on processing time and prediction error criteria. When the results are analyzed, MMPF with the proposed model selection approach; process time decreases so algorithm speed increases, there is no significant increase for prediction error. As a result, WSMS algorithm can be used effectively for maneuvering radar targets in real time follow-up.