Block sparse Bayesian learning with environmental perturbation for robust matched field processing.

Area of Science:

• Ocean acoustics
• Signal processing
• Environmental robustness

Background:

• Conventional compressive sensing matched field processing (CS-MFP) struggles with ocean environmental mismatches due to fixed prior model parameters.
• Dynamic ocean variations and parameter uncertainties limit the effectiveness of traditional CS-MFP methods.

Purpose of the Study:

• To enhance the environmental robustness of CS-MFP for improved underwater acoustic localization.
• To develop a novel CS-MFP framework that accommodates environmental perturbations and parameter uncertainties.

Main Methods:

• Constructed a block dictionary matrix using statistics of random environmental parameters.
• Proposed a multisnapshot block signal processing model for block sparse signal recovery.
• Derived an efficient multisnapshot block sparse Bayesian learning (BSBL) algorithm.
• Developed two BSBL processors with different intrablock correlation models for enhanced robustness.

Main Results:

• The proposed BSBL processors demonstrated superior robustness against environmental mismatch compared to conventional methods.
• The new methods do not require specific prior environmental parameters for effective operation.
• Validation through numerical simulations and real ocean data confirmed the enhanced performance.

Conclusions:

• The developed BSBL algorithms offer a more robust approach to matched field processing in dynamic ocean environments.
• This work advances CS-MFP techniques by effectively addressing environmental uncertainties in underwater acoustic applications.