Environmentally and statistically robust matched-field source localization based on information geometry principles.

Area of Science:

• Ocean acoustics
• Signal processing
• Geophysics

Background:

• Matched-field processing (MFP) localizes underwater sources using array and replica signals.
• Traditional MFP relies on Euclidean distance between cross-spectral density matrices (CSDMs), but is sensitive to environmental and statistical mismatches.
• Existing methods like minimum variance MFP with environmental perturbation constraints offer limited robustness.

Purpose of the Study:

• To develop a robust MFP for underwater source localization.
• To enhance robustness against environmental and statistical mismatches.
• To improve source localization resolution and reduce sidelobe levels.

Main Methods:

• Utilizing information geometry to model CS DMs as points on a Riemannian manifold.
• Quantifying CSDM similarity using geodesic distance.
• Introducing a constrained replica CSDM and employing Riemannian and modified Jensen-Shannon distances for a novel MFP.
• Developing a robust matched-field processor based on non-Euclidean distances.

Main Results:

• The proposed processors demonstrate superior robustness against environmental and statistical mismatches compared to traditional MFP.
• Simulations and experimental results confirm improved performance.
• Sidelobe levels are reduced, and resolution is enhanced.

Conclusions:

• MFP can be advanced using information geometry and non-Euclidean distances.
• The proposed robust MFP offers significant improvements for underwater source localization.
• This approach provides a more reliable method for acoustic signal processing in complex marine environments.