Bayesian optimization with Gaussian process surrogate model for source localization.

Area of Science:

• Acoustics
• Oceanography
• Signal Processing

Background:

• Accurate underwater source localization is crucial for various applications.
• Traditional methods like grid search can be computationally expensive and inefficient.
• Geoacoustic modeling requires optimizing parameters such as range and depth.

Purpose of the Study:

• To propose a sample-efficient sequential Bayesian optimization strategy for geoacoustic model parameter optimization.
• To improve the speed and efficiency of underwater acoustic source localization.

Main Methods:

• Utilizing a Gaussian process (GP) surrogate model to approximate the objective function.
• Employing a heuristic acquisition function to balance exploration and exploitation in parameter space.
• Sequentially updating the GP model with new data points.

Main Results:

• The Bayesian optimization strategy demonstrated rapid convergence to optimal solutions.
• Simulations and experimental data showed superior performance compared to grid search and quasi-random sampling.
• Effective source localization was achieved in a shallow-water waveguide environment.

Conclusions:

• Sequential Bayesian optimization offers a sample-efficient approach for acoustic source localization.
• This strategy significantly reduces the computational cost and time required for optimization.
• The method is validated for practical applications using real-world acoustic data.