Estimation of ARMA Model Order Utilizing Structural Similarity Index Algorithm
Khaled E. Al-Qawasmi · 2022 8th International Conference on Control, Decision and Information Technologies (CoDIT) · 2022
This paper comprises work on developing a novel approach for evaluating ARMA model order. The precise model order (p, q) is obtained using the structural similarity (SSIM) index when considering an ARMA (p, q) model without identifying the genuine order. From the observed data sequence, this method extracts the data covariance matrix layout. The suggested technique is based on Liang et.al. Minimum Eigenvalue (MEV) criterion. The algorithm generates a dataset of MEV covariance matrix at different Signal-to-Noise Ratios SNR to register all the aspects of this matrix; the SIM method is performed to compute the degree of similarity between the data set values and the unknown signal values to select the model which yields the most similar values. Hence, to demonstrate the significantly improved results, examples are presented.