Underwater Doppler-bearing pulse source tracking based on an outlier-tolerant variational Bayesian adaptive Kalman...

Area of Science:

• Underwater acoustics
• Signal processing
• Target tracking

Background:

• Doppler-bearing target motion analysis using sonar arrays requires accurate pulse signal parameter estimation.
• Kalman filters are crucial for this analysis but struggle with unknown noise covariance matrices and environmental uncertainties.
• Existing nonlinear variational Bayesian (VB) filters lack robustness to outliers, degrading tracking performance.

Purpose of the Study:

• To develop an outlier-tolerant nonlinear VB adaptive Kalman filter for high-precision underwater target tracking.
• To improve robustness against marine environment uncertainties and target-array geometry corruptions.
• To enhance the accuracy of measurement noise covariance matrix approximation and expectation estimation.

Main Methods:

• Proposed an outlier-tolerant nonlinear VB adaptive Kalman filter.
• Utilized a hierarchical inverse-Wishart-gamma mixture distribution model for robust outlier identification and noise covariance approximation.
• Employed adaptive high-order cubature sampling for improved expectation estimation accuracy.

Main Results:

• The proposed filter demonstrated superior performance compared to traditional methods in simulations.
• Numerical simulations showed improved tracking accuracy and convergence rates.
• Sea trials confirmed the filter's effectiveness and robustness in real-world conditions.

Conclusions:

• The developed nonlinear VB adaptive Kalman filter effectively addresses challenges in underwater target tracking.
• The combination of hierarchical mixture models and adaptive cubature sampling provides robust and accurate tracking.
• The filter offers a significant advancement for high-precision underwater non-cooperative target tracking applications.