Consistency analysis of a frequency domain subspace algorithm for multi-component harmonic retrieval

A new frequency domain algorithm for multi-component harmonic retrieval has been developed recently. The method is based on using discrete Fourier transform (DFT) data and enables the user to select frequency sub-bands within which the frequencies of the harmonics of interest are known to reside. The technique provides a very good attenuation of colored noise and other disturbances residing outside the frequency band of interest. The paper presents a consistency analysis of the subspace based estimation algorithm. The analysis provides sufficient conditions for consistency of the sinusoidal frequency estimates in terms of required number of DFT data. A small numerical example is also presented comparing the new approach with the well known ESPRIT method. The result clearly indicates that the method has a large potential.