Characterization of windowing effects in adaptive extrapolation of sinusoids

The adaptive weighted norm extrapolation (AWNE) method is well suited for spectral analysis of sinusoidal signals from short data records. The time-domain shape of the envelope of the result is considered to be a windowing on the original signal. This paper uses statistical methods and harmonic random processes to estimate this imposed window. The squared Tukey window is found to closely match the average window shape. The windowing assumption is then evaluated using signal comparison criteria. Finally, the application of the window model to SAR imaging is illustrated.