Wide-band Direction of Arrival (DOA) estimation with sensor arrays is anessential task in sonar, radar, acoustics, biomedical and multimediaapplications. Many state of the art wide-band DOA estimators coherently processfrequency binned array outputs by approximate Maximum Likelihood, WeightedSubspace Fitting or focusing techniques. This paper shows that bin signalsobtained by filter-bank approaches do not obey the finite rank narrow-bandarray model, because spectral leakage and the change of the array response withfrequency within the bin create \emph{ghost sources} dependent on theparticular realization of the source process. Therefore, existing DOAestimators based on binning cannot claim consistency even with the perfectknowledge of the array response. In this work, a more realistic array modelwith a finite length of the sensor impulse responses is assumed, which stillhas finite rank under a space-time formulation. It is shown that signalsubspaces at arbitrary frequencies can be consistently recovered under mildconditions by applying MUSIC-type (ST-MUSIC) estimators to the dominanteigenvectors of the wide-band space-time sensor cross-correlation matrix. Anovel Maximum Likelihood based ST-MUSIC subspace estimate is developed in orderto recover consistency. The number of sources active at each frequency areestimated by Information Theoretic Criteria. The sample ST-MUSIC subspaces canbe fed to any subspace fitting DOA estimator at single or multiple frequencies.Simulations confirm that the new technique clearly outperforms binningapproaches at sufficiently high signal to noise ratio, when model mismatchesexceed the noise floor.
展开▼
