On-line tuning methods for adapting a Kalman filter to unknown noise statistics
William L. Brogan, R.L. Olson · 2003
Frequently, the noise statistics required for implementing the Kalman filter are not precisely known. Several candidate time-average statistical measures of a tuned filter and a number of adaptive adjustment schemes for achieving the tuned goal are explored. A promising method adapted from the Jayant one-bit equalizer is presented.>