Abstract
This paper focuses on the parameter estimation problem of multivariate output-error autoregressive moving average (M-OEARMA) systems. By applying the auxiliary model identification idea and the decomposition technique, we derive a two-stage recursive least squares algorithm for estimating the M-OEARMA system. Compared with the auxiliary model based recursive least squares algorithm, the proposed algorithm possesses higher identification accuracy. The simulation results confirm the effectiveness of the proposed algorithm.