Adaptive and compressive matched field processing.

Area of Science:

• Acoustics
• Signal Processing
• Optimization

Background:

• Matched field processing (MFP) is a beamforming technique for source localization using array data and replica vectors.
• Traditional MFP methods often result in sparse solutions due to fewer sources than replicas.

Purpose of the Study:

• To reformulate the matched field problem using compressive sensing (CS) implemented with basis pursuit.
• To evaluate the performance of CS in source localization, particularly its robustness and comparison to existing methods.

Main Methods:

• Utilized compressive sensing (CS) and basis pursuit to transform the MFP into an underdetermined convex optimization problem.
• Applied a row-sparsity constraint to estimate unknown source amplitudes and select optimal replicas from the dictionary.
• Compared CS performance against Bartlett and adaptive white noise constraint processors using simulations and experimental data.

Main Results:

• CS performance is equivalent to the Bartlett processor for a single source across any number of snapshots.
• CS demonstrates modest localization performance improvement over the adaptive white noise constraint processor for incoherent sources.
• CS accommodates coherent sources, unlike most adaptive processors, and shows robustness to data-replica mismatch.

Conclusions:

• Compressive sensing provides a powerful and robust approach to matched field processing for source localization.
• CS offers advantages in handling coherent sources and exhibits improved performance and reduced ambiguity compared to conventional methods.