Rank-One Extensions of the Generalized Hermitian Eigenvalue Problem for Adaptive High Resolution Array Processing

The determination of number, location, and movement of radiating sources using passive arrays of sensors is an important problem in radar imaging, medical imaging, and seismology. High resolution array processing, methods utilize spectral information contained in the observed signal correlation matrix in order to decompose incoming signals into signal and noise components. This allows extraction of various characteristic parameters such as source intensity and direction of arrival. Treatment of nonstationary signals in a colored noise environment necessitate the, analysis of generalized Hermitian pencils. In order to conserve computational resources and permit real time processing of information, one may wish to process these matrix pencils adaptively only up to the smallest order necessary for full estimation of the desired signal parameters.