On bias in nonlinear (Kalman) filtering

Author

Eitelberg, E. ; Heystek, L.

Author_Institution

UEC Projects, Durban, South Africa

fYear

1989

fDate

32682

Firstpage

31

Lastpage

36

Abstract

The practical problem of passive range tracking of a moving source is exploited to demonstrate how to eliminate bias in nonlinear (Kalman) filtering. The approach is to formulate the measurements in an indirect linear form and filter only linear (dynamic) relationships by `shifting´ all nonlinearities into additive system disturbance terms. Specifically, a nonlinear Kalman filter is developed for passively tracking the range of a moving acoustic source. Simulation results are presented and confirm the theoretical predictions

Keywords

Kalman filters; acoustic signal processing; correlation methods; filtering and prediction theory; additive system disturbance terms; bias eliminations; cross correlation methods; filter only linear dynamic relationships; indirect linear form; moving acoustic source; nonlinear Kalman filter; nonlinearities shifting; passive range tracking; time delay measurements; Acoustic measurements; Acoustic sensors; Delay effects; Delay estimation; Equations; Information filtering; Information filters; Kalman filters; Measurement errors; Time measurement;

fLanguage

English

Publisher

ieee

Conference_Titel

Communications and Signal Processing, 1989. COMSIG 1989. Proceedings., Southern African Conference on

Conference_Location

Stellenbosch

Print_ISBN

0-87942-713-2

Type

conf

DOI

10.1109/COMSIG.1989.129012

Filename

129012