Identification of multivariable ARMA models by use of fast algorithms

This paper describes two procedures for identifying the parameters of a multivariable AX! model by use of the auto correlation function of the signal. The procedures consist in fitting a pure AR model or a pure MA model to data generated bv the ANLI model. The parameters of these AR or MA models are indentified by use of now well-known fast algorithms and the ARMA parameters are then computed by solving simple algebraic equations. max