Robust high-resolution eigenspace DOA estimation

A subspace approach to direction-of-arrival (DOA) estimation that incorporates high resolution with robustness to array response uncertainties is presented. It is argued that one-dimensional search algorithms can be made more robust to array errors if the errors can be characterized in terms of a structured distribution in the observation space. This argument leads to a discussion of methods for low-rank and full-rank perturbations. In both cases finite sample effects and model errors are simultaneously dealt with by generalizing the high resolution subspace approach of CLOSEST. A simulation study is included to compare the three robust methods introduced with MUSIC, and a previously proposed robust weighted version of MUSIC, for cases of low and full rank model perturbations