Robust DOA Tracking of an Underwater Target in Non-Gaussian and Nonstationary Environmental Noise

Estimating the direction of arrival (DOA) of underwater targets using hydrophone array signals is a key research issue in the field of underwater acoustic signal processing. The strong nonlinearity of array signal models, the high-dimensionality of multisnapshot sonar acoustic vector (SAV) measurements, and the non-Gaussian and nonstationary noise in complex underwater environments make continuous and robust DOA tracking a significant challenge. To address this challenge, this article proposes a robust DOA tracking algorithm for underwater targets in non-Gaussian and nonstationary noise environments. First, by applying the central limit theorem, the high-dimensional non-Gaussian SAV measurements are reduced to a 1-D innovation measurement that, under sufficient sample conditions, can be approximated as a Gaussian distribution. This Gaussian innovation serves as the measurement input for the filtering process, thereby simplifying the treatment of complex, high-dimensional, and non-Gaussian noise in the SAV-based DOA tracking algorithm. Next, a variational Bayesian method with multiprior constraints dynamically infers innovation measurement distribution parameter to handle nonstationarity. Finally, particle swarm optimization is applied to mitigate particle depletion and enhance filtering performance. Simulations and South China Sea data (July 2021) validate that the algorithm significantly outperforms conventional methods in robustness and accuracy under complex noise conditions.