Utilisation of the MV spectrum for diagnosis and estimation of harmonic fields

The authors consider the identification of a single channel multidimensional harmonic signal field contained in an unknown and possibly nonstationary noise field. A technique is presented using the convergence-based point spectra developed by Frazho and Sherman (1990, 1991). This technique has the advantage of discerning between point and narrowband noise spectra based on the monotonic convergence properties of a sequence of minimum variance (MV) spectra. Two examples are presented to demonstrate the properties and utility of the approach, also a comparison to the multidimensional discrete Fourier transform and autoregressive modelling techniques