Title :
Nonlinear filtering with small observation noise
Author_Institution :
Div. of Appl. Math., Brown Univ., Providence, RI, USA
Abstract :
A study is made of the nonlinear filtering of diffusions when the observation noise covariance is proportional to ε2, a small parameter. It is shown that the suboptimal solution obtained by the extended Kalman filter is an approximation in the L2 sense of order ε2 to the best nonlinear filter. The technique involves the use of an efficient linearization method obtained via the Girsanov transformation
Keywords :
diffusion; filtering and prediction theory; linearisation techniques; Girsanov transformation; diffusions; extended Kalman filter; linearization method; nonlinear filtering; observation noise covariance; Computer errors; Eigenvalues and eigenfunctions; Error analysis; Filtering; Filters; Gaussian processes; Mathematics; Probability; Tellurium; Tiles;
Conference_Titel :
Decision and Control, 1989., Proceedings of the 28th IEEE Conference on
Conference_Location :
Tampa, FL
DOI :
10.1109/CDC.1989.70642
