Optimal minimal-order least-squares estimators via the general two-stage Kalman filter

Abstract

A direct derivation of the optimal minimal-order least-squares estimator (OMOLSE) is presented using the recently developed general two-stage Kalman filter (GTSKF). Using this new result, the reduced-order estimators of O'Reilly and Fairman are readily shown to be equivalent. A practical implementation issue to consider these two estimators is also addressed.