Fourth-order cumulants based method for continuous-time EIV fractional model identification

The errors-in-variables (EIV) system identification problem concerns the dynamic systems whose discrete input and output are corrupted by additive noises, that can be white, colored and/or mutually correlated. In this paper, a new estimator based on Higher-Order Statistics (fourth-order cumulants) is proposed for continuous-time system identification with fractional models. Under some assumptions on the distributional properties of the noise and noisefree signals, the fractional fourth-order cumulants based least squares (ffocls) estimator gives consistent results. A numerical example illustrates the performance of the proposed method.