Direction of arrival estimation in mixed compound-Gaussian and Gaussian noise
Olivier Besson, Yuri I. Abramovich, Ben A. Johnson · 2016
Parameter estimation of directions of arrival (DoA) of multiple signals received in the presence of a combination of a strong non-Gaussian external noise and weak internal white Gaussian noise represents a physically relevant mixed noise circumstance, yet one where in very heavy-tailed distributions, theoretical performance bounds do not exist. We suggest a practical performance bound by examining a related model where the texture of the compound-Gaussian component treated as an unknown and deterministic quantity is estimated together with DoAs and a basis of the signal subspace. A lower bound for DoA estimation, based on known texture values, is derived and a prediction of the actual mean-square error is presented, based on separation between external/internal-noise dominated samples. Numerical simulations indicate that a suggested iterative DoA estimation technique performs close to the bound and has estimation accuracy that improves upon a number of existing routines.