Nonstationary parametric system identification using higher-order statistics

Abstract

In this paper, consideration is given to the estimation of the parameters of a time-varying linear model. It is shown that the time-varying ARMA model of single-input single-output (SISO) system is equivalent to the time-invariant ARMA model of multi-input multi-output (MIMO) system. Novel methods for the parameter estimation task are developed based on the concepts of higher order statistics (HOS). The proposed algorithms are compared with a range of existing (second order) algorithms via simulation studies which cover several systems at various signal to noise ratios (SNRs). Through these studies, the robustness of the HOS based algorithms to additive Gaussian noise is demonstrated.