Adaptive MA-parameter estimation with modulated cumulants
Thomas Kaiser · 2003
In this paper we propose a new algorithm for adaptively estimating the parameters of a time variant MA-model based on so-called modulated second- and third-order cumulants. The concept of modulated cumulants enables the use of all available statistical information up to third-order and therefore the new algorithm does not only lead to a faster convergence but also to higher parameter estimation accuracy compared to known methods. We will illustrate this behaviour by numerical examples.