SINGULAR-VALUE DECOMPOSITION-BASED MA ORDER DETERMINATION OF NON-GAUSSIAN ARMA MODELS

This paper addresses the singular value decomposition (SVD) based MA order determination of non-Gaussian processes using higher order statistics. It is shown that the moving average (MA) order determination of ARMA models is equivalent to the rank determination of a certain error matrix, and the SVD-1 approach is proposed. Its simplified form is referred to as the SVD-2 which is applied to pure MA models. To improve the robustness of the order selection, we suggest a combination of the SVD and the product of diagonal entries (PODE) test. Some interesting applications of the two SVD approaches are presented. Simulations verify the performance of our two approaches.