Harmonic retrieval using cumulant-based estimator for ARMA systems

Abstract

Guog U~versity, Taoyuas T i m R0.C. Email: d8692IM8@ntt~eduhv and ac@cht.com.nv ABSTRACT -In this paper, we propose a cumulant-based estimator for ARMA systems. The polyspectra, cumulants, and various other related statistics, such as bicepstra and bicoherence are all used to develop cumulant-based algorithms for estimating the parameters of linear (e.g. ARMA) or nonlinear processes. The use of cumulant-based estimator is useful: (1) if the additive noise is Gaussian and the signal is non-Gaussiaa, (2) the linear system is non-minimum phase (that is, mixed-phase), or (3) the process is nonlinear. 'he singula~ value plot show several significant singular values comsponding to me quadratically coupled triplet; as in the case of the power specbum (harmonics in noise), overestimating the number of harmonics usuaUy leads to bener results.