Assessment of influence of systematic errors on the precision with which the normalized frequency of a sinusoidal signal is determined by means of a discrete fourier transformation with interpolation

Abstract

The influence of systematic errors caused by the infleunce of the imaginary part in the spectrum of a sinusoidal signal on the estimator of its normalized frequency is analyzed by means of a discrete Fourier transformation with interpolation and a window with maximal rate of descent of the side lobes. An expression for the absolute error of the normalized frequency is presented and a condition for the minimal integral number of cycles of the sine curve is found; note that computation of the actual integral number of cycles enables us to assure that this error will be less than some specific value. The reliability of the expressions that are obtained is confirmed by a computer simulation.