Harmonic retrieval in colored non-Gaussian noise using cumulant and autocorrelation

Abstract

This paper proposes a higher order cumulant-based approach to harmonic retrieval in non-Gaussian ARMA noise. Hilbert transform is used to transform the real measurements into their complex counterpart, then a kind of elaborately defined fourth-order cumulant is employed to identify the AR parameters of the non-Gaussian noise. After prefiltering the noisy measurements with the identified AR polynomial, SVD-TLS method can be applied to retrieve harmonics. This method releases the asymmetrically-distributed assumption of the non-Gaussian noise and is also efficient when there is quadratic phase coupling in harmonics. Simulation examples are presented to support the result of this paper.