Joint towed array shape and direction of arrivals estimation using sparse Bayesian learning during maneuvering

Area of Science:

• Underwater acoustics
• Signal processing
• Array processing

Background:

• Large aperture towed arrays are crucial for detecting weak underwater targets.
• Array configuration errors during maneuvering degrade beamformer performance.
• Current methods for estimating array positions using sensors or acoustic sources are often inadequate.

Purpose of the Study:

• To propose a novel algorithm, adaptive bow sparse Bayesian learning (ABSBL), for jointly estimating towed array configuration and signal directions of arrival (DOAs).
• To address the limitations of existing methods in accurately determining array element positions during maneuvering.

Main Methods:

• Developed an adaptive bow (AB) sparse Bayesian learning (SBL) algorithm (ABSBL).
• Modeled the towed array's shape as a parabola during slow turns.
• Treated the array bow as a hyperparameter within the SBL framework for joint estimation.
• Utilized received acoustic data for estimation.

Main Results:

• Simulations demonstrated ABSBL's accuracy in estimating array bow and target DOAs when turning direction is known.
• Application to MAPEX2000 data showed agreement between ABSBL estimates and those from relative time delays.
• ABSBL performance surpassed GPS-based methods using the water-pulley model.

Conclusions:

• The ABSBL algorithm provides accurate joint estimation of array bow and DOAs from acoustic data.
• This method offers a potential alternative to engineering sensors for array configuration estimation.
• The approach shows promise for improving underwater acoustic detection systems.