Adaptive frequency estimation for sinusoidal signals in colored non-Gaussian noises (Q5928980)

This paper deals with the estimation of harmonic signal components in colored, additive noises which are described by a non-Gaussian ARMA process. The sinusoidal frequencies at each time instant are the aim of the estimation process. For this purpose, an adaptive scheme is developed that is utilizing certain properties of cumulants. This scheme is called AFE-23 because it is based on the second- and third-order statistics of the measurement process. The author presents a convergence analysis as well as computer simulations which are comparing the AFE-23 algorithm to a correlation-based approach. In the experiments, the new algorithm is tracking the frequency changes correctly.