An approach to DOA estimation of wide-band sources based on AR signal modeling

In this paper we propose a wide-band direction-of-arrival estimation method that relies in the assumption that the sources are autoregressive. A matrix formulation is introduced in the frequency domain so that all available information from the sources can be incorporated thus allowing a new insight in the estimation process. We show that it is possible to deal with different sources regardless the closeness of their center frequency and their relative location. By incorporating the AR coefficients of each source the scope of possibilities is opened. This allows to resolve close sources with quite a few number of sensors. Indeed, bearing estimations can be done with fewer sensors than sources. In the experiments we explore the potential applications of our method in real situations where it is needed to resolve close sources that share the same frequency band and the number of sensors at disposal is low.