New algorithms of extrapolation and spectral estimation for discrete band-limited signals based on Fourier transform

摘要

Two algorithms of extrapolation and spectral estimation for discrete bandlimited signals are proposed that are based on Fourier transform. The two algorithms, the matrix algorithm and the iterative FFT algorithm, can provide more accurate spectral estimates and better extrapolation performance than other algorithms developed for discrete bandlimited signals with a smooth spectrum. Numerical examples show that the iterative FFT algorithm, which uses accelerated convergence, is a robust extrapolation algorithm for the noisy data.