Performance Analysis of Multiple Kalman Filter Systems

Various Kalman Filters have been used in literature to remodel noise and follow a target in Cartesian coordinates whose measurements come in polar coordinates. In addition, since the target to be tracked in real life is not bound to a single motion model, systems enabling the interaction of multiple motion models have been used. In this article, these filters and their motion models are defined and the performance of these filters and multiple filter systems are evaluated by simulations with various scenarios. In this simulations, all types of Kalman Filters show similar performances, but DCMKF (Debiased Consistent Converted Measurement Kalman Filter) is faster than others. Among multiple filters systems, IMM (Interacting Multiple Model) outperforms CUSUM(Cumulative Sum).