FREQUENCY ESTIMATION BY IDFT AND QUANTIZATION NOISE

D. Agrez

Abstract

Possibilities of error reductions of the frequency estimation by the multipoint interpolated discrete Fourier transform (IDFT) for rectangular and von Hann windows are described. An analysis is made to study the influence of the timedomain noise on the results of a DFT and interpolations. The quantization noise influence on the amplitude coefficients of DFT is in first approximation modeled by white noise in the time domain and Gaussian noise in the frequency domain. The noise influences on estimations are increased with the root of a number of the interpolation points, that is with apparent effective narrowing of the tapered window. Interpolations with larger number of points decrease the systematic errors and increase the noise distortion of results, with smaller number of points there is an inverse appearance. Increasing a number of the used DFT coefficients is reasonable until the systematic error drops under the noise floor.

Keywords

frequency estimation, IDFT, quantization noise

Download

IMEKO-WC-2000-UNC-P493.pdf

DOI

-

Back to the proceedings