Write a MATLAB program to estimate the order of an all-pole random process using the Akaike final prediction error, the minimum descriptor length, the Akaike information criterion, and Parzen's CAT. Evaluate the accuracy of the estimates that are produced with these methods if the modeling errors εp are derived using the autocorrelation method. Consider both broadband and narrowband AR processes. How much do the estimates change if εp is replaced with the Burg error,εBP? Discuss your findings.