Robust wavefield interpolation for adaptive wideband beamforming

摘要

In this paper we consider the application of interpolation-based focusing for wideband array processing and direction finding using wavefield interpolation matrices. The focusing approach exploits the wideband characteristic of the signals for coherent subspace processing, achieving improved performance and reduced computational complexity. We investigate the use of wavefield modeling-based frequency transformations of the array manifold, which are data independent and do not require any initial direction of arrival (DOA) estimates. We apply these transformations to the array wideband data, constructing a virtual wavefield-interpolated narrowband-generated subspace (WINGS), on which any narrowband adaptive beamforming scheme may be implemented. WINGS transformations are also shown to be optimal for frequency-invariant beamforming (FIB). We treat the important issue of robust WINGS processing, i.e. reducing the transformation's sensitivity to errors in the array manifold, such as sensor gain, phase and location errors. We develop two novel robust forms of the WINGS transformations designed to ensure robustness by controlling the noise gain of the transformation, and study their performance numerically. Finally, we demonstrate via simulation the effectiveness of WINGS processing for minimum variance distortionless response (MVDR) beamforming.