Title :
MA-model identification using modulated cumulants
Author_Institution :
Dept. of Commun. Eng., Duisburg Univ., Germany
Abstract :
In this paper we present a new linear method for estimating the parameters of a moving average model from modulated cumulants of the observations of the system output. The input sequence must be non-Gaussian with some special properties described in the text. Both recursive closed-form and batch least-squares versions of the parameter estimator are presented. The proposed linear method utilizes a complete set of the relevant output statistics, so it should lead to more accurate parameter estimates compared to other linear methods. This property is illustrated through simulations. Furthermore it uses two different cumulants of arbitrary order and is therefore not restricted to the second and third order case
Keywords :
higher order statistics; least squares approximations; modulation; moving average processes; parameter estimation; recursive estimation; signal processing; MA-model identification; batch least-squares estimator; linear method; modulated cumulants; moving average model; non-Gaussian input sequence; output statistics; parameter estimation; recursive closed-form estimator; simulations; system output; Artificial intelligence; Equations; Hydrogen; Parameter estimation; Random processes; Random variables; Recursive estimation; Statistics;
Conference_Titel :
Digital Signal Processing Workshop, 1994., 1994 Sixth IEEE
Conference_Location :
Yosemite National Park, CA
Print_ISBN :
0-7803-1948-6
DOI :
10.1109/DSP.1994.379853