The autocorrelated noise filtering problem: the ISMC filter in a specific case of distance measuring

Abstract

In a previous paper we were working on a electronic travel aid for blind people based on infrared sensors. The signals coming from them are affected by a great noise that also with the use of low pass filter cannot be clean well. Motivated by the improvement of the system, in this paper we show a novelty way to filter autocorrelated noise based on a probabilistic description of the process. We apply an indexed semi-Markov model in order to filter the signal coming from the infrared sensor. We conduce first of all a data analysis on the noise in order to understand well its form. We give the general formulation of the new ISMC filter and at last we compare the results with a particular kind of Kalman filter for the specific stochastic application.