Title :
An empirical formula for selection of AR model order
Author_Institution :
North Electron. Equip. Inst., Beijing, China
Abstract :
The paper presents the concise principle of power spectral estimation using the AR model (AR-PSE). The AR-PSE needs a priori knowledge of the order M of the AR model. An empirical formula for the AR model order is derived. Examples are given of the formula applied in practical signal analysis. The results testify that the empirical formula for the selection of the AR model order for AR-PSE is valid for real signals. It can be concluded that, so long as the resolution requirement is satisfied, the lower the order M is, the better the performance of the AR-PSE is.
Keywords :
autoregressive processes; parameter estimation; signal resolution; spectral analysis; AR model order; empirical formula; power spectral estimation; signal analysis; signal resolution; Equations;
Conference_Titel :
Antennas and Propagation Society International Symposium, 2003. IEEE
Conference_Location :
Columbus, OH, USA
Print_ISBN :
0-7803-7846-6
DOI :
10.1109/APS.2003.1217410