Autocorrelation-based algorithm for ARMA model order selection in colored gaussian noise

In this paper, we have addressed the ARMA model order selection problem for the case of colored Gaussian noise using autocorrelation. The most well known solutions for the ARMA model order problem are the Akiake information criterion (AIC), the minimum description length (MDL), and the minimum eigenvalue (MEV) criterion. In the MEV method, observation and/or modeling error is assumed to be zero-mean while Gaussian. This paper presents a generalization of the original results in the MEV method to the colored Gaussian noise for the second order statistics. Simulations show the performance of the generalization results