Normalized ARMA Schur algorithm

A Schur-type algorithm for estimating the one-dimensional autoregressive moving average (ARMA) model from the correlations of the observed inputs and outputs of an unknown system is proposed. The algorithm represents a faster alternative to conventional methods. It is characterized by flexibility in the arrangement of order-update operations, and the simplicity and modularity of the computational architecture. The substantial amount of parallelism contained in the algorithm makes it suitable for VLSI implementation.<<ETX>>