Analysis of an adaptive extrapolation algorithm on the recovery of harmonic signals

Abstract

The adaptive weighted norm extrapolation (AWNE) method is well suited for recovering sinusoidal signals from short data records since they have well concentrated spectral representations. This paper analyzes the performance of the algorithm in the recovery of sinusoidal signals; validates the assumption that the result of the algorithm is a windowed version of the original; and derives a form for this imposed window. The role of the autocorrelation of the window used in the AWNE algorithm plays a key role in our analysis. We also provide a new interpretation of the AWNE method as a best subspace selection method for signal approximation.