ARMA MODEL ORDER ESTIMATION BASED ON THE EIGENVALUES OF THE COVARIANCE-MATRIX

Much research has focused on the problem of estimating the model order of ARMA processes. The most well-known of the proposed solutions for this problem include the final prediction error (FPE), Akaike Informat ion Criterion (AIC), and minimum description length (MDL). In this pap er a new approach for model order determination based on the MDL crite rion is proposed and shown to depend on the minimum eigenvalues of a c ovariance matrix derived from the observed data. As a result, a new se lection procedure for estimating the model order via MDL is proposed. Examples are given that illustrate the significantly improved accuracy of the proposed technique.