DOA estimation under Bernoulli-Gaussian impulsive noise

Abstract

Abstract Direction of arrival algorithms have ability to distinguish between directions of individual sources under additive white Gaussian noise. However, presence of the impulsive noise can corrupt the angles estimation. In this paper, the performance of MUSIC and ESPRIT, as estimation algorithms, is investigated in impulsive noise with white Gaussian noise environment. The impulsive noise is modelled using Bernoulli-Gaussian model and the effect of varying its statistical parameters on the estimation ability is tested. Three sources with different angles are simulated and a uniform linear array is utilized to capture the impinging signals. The performance analysis includes impulsive noise probability of occurrence and amplitude intensity which both control the impulsive power. The mean square error of the estimated angles is adopted as a metric to measure the performance. The simulation results provide a map of all the expected possible impulsive noise cases which influence the MUSIC and ESPRIT direction of arrival algorithms. The expected cases show that the performance degradation happening in ESPRIT is more than that of MUSIC. At 0 dB SNR, the impulsive noise increases the mean square error by almost 15 dB and 20 dB for MUSIC and ESPRIT, respectively.