Sliding DFT and Zero Padding

Abstract

In many circumstances, it is preferred to simultaneously use a recursive Discrete Fourier Transform (DFT) and the zero padding, the former for its computational efficiency, the latter for the interpolation in the frequency domain. In the present article, we develop a new sliding DFT algorithm, which performs recursive spectral estimation of zero-padded time series. In order to ensure the stability of the process, we selected the estimator with guaranteed stability through the damping factor. Error analysis shows that the damping factor produces systematic errors and the errors in question are dependent upon the phase of the input signal.