Title :
Parameter estimation for linear alpha-stable processes
Author :
Swami, Ananthram ; Sadler, Brian
Author_Institution :
Army Res. Lab., MD, USA
Abstract :
Although alpha-stable processes have infinite variance, one can define and consistently estimate the normalized correlations and cumulants of linear processes with stable innovations. Hence, conventional techniques can be used to estimate the parameters of nonminimum phase alpha-stable processes.
Keywords :
autoregressive moving average processes; correlation methods; higher order statistics; linear systems; parameter estimation; signal sampling; stability; ARMA; cumulants; impulse response; infinite variance; linear alpha-stable processes; nonminimum phase alpha-stable processes; normalized correlations; normalized fourth-order moments; parameter estimation; sample estimators; stable innovations; Additive noise; Autoregressive processes; Closed-form solution; Convergence; Maximum likelihood estimation; Parameter estimation; Phase estimation; System identification; Technological innovation;
Journal_Title :
Signal Processing Letters, IEEE
