Chapter 1: Invited Addresses and Tutorials on Signals, Coding,
Systems and Intelligent Techniques
Title: MV AR Model Order Identification Using Partitioning Theory
Author(s): Stylianos Sp. Pappas, Assimakis K. Leros, Sokratis K. Katsikas
Address: Department of Information and Communication Systems Engineering, University of the Aegean, 83200, Karlovassi, Samos, Greece | Department of Information and Communication Systems Engineering, University of the Aegean, 83200, Karlovassi, Samos, Greece | Department of Information and Communication Systems Engineering, University of the Aegean, 83200, Karlovassi, Samos, Greece
Reference: 12th International Workshop on Systems, Signals and Image Processing pp. 51 - 55
Abstract/Summary: In this paper the multimodel partitioning theory (Lainiotis 1976a; 1976b; 1971) is applied for simultaneous order identification and parameter estimation of multivariate (MV) autoregressive (AR) models. The simulations show that the proposed meted succeeds to select the correct MV AR model order and estimates the parameters accurately in very few steps and even with a small sample size. The results are compared to many other well established order selection criteria, namely Akaike's Information Criterion (1969), Schwarz's Bayesian Information Criterion (1978), Hannan's and Quinn’s (1979), Brockwell's and Davis' (1991) and C. C. Chen, R. A. Davis and P. J. Brockwell order determination criteria (1996). Finally as an extension it will be shown that the algorithm is also successful in tracking model order changes in real time.
Order a copy of this article