Optimal and Suboptimal State Estimation for Linear Systems with Additive Cauchy Noises

摘要

Only few estimation methods can converge in the presence of impulsivemeasurement and process noises without the use of augmentedheuristic schemes. To understand the numerical behavior of theseschemes, the performance of the optimal Idan/Speyer Cauchy Estimator(ISCE) is compared with those of the particle filter (PF) and Gaussiansum filter (GSF) while allowing similar computation times for the differentestimators. In addition, the number of particles at each step forthe PF and the number of Gaussian components kept at each step forthe GSF are increased and their performance relative to the ISCE isnumerically studied for a scalar and a two-state dynamic system.