Exponential data weighting in the Kalman-Bucy filter

Abstract

The use of exponential weighting of data (favoring more recent data) is considered for continuous-time and discrete-time Kalman-Bucy filters. Quantitative consideration is given to matters such as the computation, stability and stationarity of filters designed with data weighting. The problem of divergence or data saturation is explored, and the use of exponential weighting is compared with other techniques for dealing with the problem. Show more