The global matched filter: application to DOA estimation with an uniform circular array

The problem of estimating the direction of arrival (DOA) of narrow-band sources impinging on a uniform circular array is considered. We present a method that uses as input the values of a small number of uniformly spaced beams and apply a model-fitting approach taking into account the statistical properties of the beams. The approach called the "global matched filter" fits simultaneously to the observations all the elements from a given dictionary needed to explain them. It chooses among all the so-obtained representations satisfying a given physically appealing constraint the one with minimal energy. This representation is in general quite sparse. The method drastically improves upon the conventional beamformer and has performance comparable to the best high resolution techniques. It further applies when the number of sources exceeds the number of sensors, a situation that cannot be handled by standard high resolution techniques.