An order recursive generalized least squares algorithm for system identification

In system identification an order recursive algorithm is of practical interest. However, such an algorithm, for obtaining asymptotically unbiased and consistent parameter estimates in a noisy linear system is lacking. An order recursive generalized least squares (GLS) algorithm is presented. This is based on a new recursive method for computing the least squares inverse developed in Section II, and is an extension of Hsia's GLS algorithm [2].