Alternative state estimation from nongaussian observation

An alternative state estimation scheme to extract gaussian message from nongaussian observation was proposed. The filter consists of a modification of Kalman filter to include a prefilter instead Kalman gain scheme. This prefilter is a conection in paralell of three branches: the first one, constant, has certain control over the transient; the second one, is basically a sum operator that “gaussianizes” the output estimation error; and a third one, a difference operator, is sensible to output estimation error bias. The three branches give an aspect of the PID structure on the prefiltering of the innovation process. An analysis, by simulation, and comparison with other structures is presented.