Reduced Rank TLS Array Interpolation for DOA Estimation

Abstract:
Important array signal processing techniques such as Spatial Smoothing, Forward Backward Averaging and Root-MUSIC require arrays with precise and specific geometries and responses. However, building sensor arrays with such demanding characteristics is not always possible. To deal with these possible limitations the real array response can be interpolated into the desired response applying array interpolation methods. In this work we study array interpolation methods for cases where the knowledge of the real array response is incomplete or contains errors. To address these imperfections a novel Total Least Squares (TLS) approach for calculating the transformation matrices is presented. Furthermore, a novel reduced rank regression approach is used to reduce the bias introduced by the transformation matrix onto the final direction of arrival (DOA) estimation.