Off-Grid Bayesian Algorithm for Near-Field Underwater Acoustic Source Localization

In this paper, we address the problem of localizing unknown underwater acoustic sources inside a boat using signals captured by a sensor array. The near-field nature of this scenario prevents the problem from being modeled as a direction-of-arrival estimation. In this context, the source localization accuracy is significantly limited at low frequencies, considering the requirement of coarse source grid. To overcome this limitation, we propose an off-grid extension to a Bayesian algorithm, allowing source positions to be estimated beyond the predefined discretized grid. Numerical simulations demonstrate that our approach effectively improves source localization accuracy. Furthermore, our method enhances signal level differentiation between true source positions and other points, making it particularly attractive in high- noise environments.