Curve fitting to asymmetrical chromatograms by the extended Kalman filter in frequency domain

An algorithm is presented to perform curve fitting on the basis of the Fourier transform of a signal. The curve fitting has been done with the exponentially modified Gaussian model, obtained when convolving a Gaussian distribution and an exponential decay function. The parameter estimation has been carried out by means of the two-dimensional extended Kalman filter. The goodness-of-fit has been characterized by the trace of the covariance matrix of the estimated parameters. The precision of the estimated parameters obtained by fitting either in time or frequency domain is compared.