Reduced rank processing for oversampled arrays

In this paper, we study array processing techniques for oversampled passive arrays. Oversampling occurs when an array contains sensors that are closely spaced relative to the operational wavelength, or when the sources are known a priori to be confined to a limited angular range. We examine here the application of the recently introduced reduced rank processing (RRP) approach to several widely used array processing methods for source detection and bearing estimation, such as maximum likelihood and MUSIC, and show how, when an array is oversampled, RRP can be used to reduce their computational load. Furthermore, we use asymptotic performance analysis to show that RRP for oversampled arrays can significantly enhance the performance of signal-subspace-based methods.