High resolution estimation for sub-gaussian stable signals in a linear array model

The authors describe a high resolution subspace fitting (HRE) algorithm for robust estimation of direction of arrival (DOA) in the uniform linear array model with heavy-tailed signals and noise. Electromagnetic disturbances on telephone lines, atmospheric noise and underwater acoustic noise often exhibit heavy-tailed behaviour with differing characteristics. Although statistical models under Gaussian assumptions of signals and noise have been extensively investigated in the literature, there is limited research on robust methods in the non-Gaussian setting. A general model with sub-Gaussian alpha-stable signals is described, which includes the isotropic alpha-stable, and independent and dependent Gaussian models as special cases. It is shown that the HRE algorithm provides strongly consistent estimates of the DOAs. In addition, Monte Carlo simulation studies show that the proposed algorithm works extremely well for closely spaced targets, and outperforms the multiple signal classification-type algorithms for strongly dependent signals, both in the stable and the Gaussian cases