Underdetermined Wideband DOA Estimation for Off-Grid Sources with Coprime Array Using Sparse Bayesian Learning.

Area of Science:

• Signal Processing
• Array Signal Processing
• Statistical Inference

Background:

• Direction of Arrival (DOA) estimation is crucial for sensor array applications.
• Underdetermined DOA estimation with coprime arrays offers enhanced degrees of freedom (DOFs).
• Sparse-based DOA methods face challenges with dictionary mismatch and off-grid sources.

Purpose of the Study:

• To apply Sparse Bayesian Learning (SBL) for underdetermined wideband DOA estimation using coprime arrays.
• To address the dictionary mismatch problem inherent in sparse DOA estimation.
• To evaluate the performance of SBL against other state-of-the-art sparse DOA algorithms.

Main Methods:

• Utilized coprime arrays with an augmented covariance matrix to increase DOFs.
• Employed SBL with fixed-point updates to resolve the dictionary mismatch problem.
• Leveraged SBL's ability to automatically select sparsity and approximate non-convex optimization.

Main Results:

• SBL demonstrated effective resolution of underdetermined wideband DOA estimation with coprime arrays.
• The proposed SBL method significantly improved detection and estimation performance.
• SBL outperformed LASSO, SOMP-LS, SOMP-TLS, and OGSBI in numerical simulations.

Conclusions:

• SBL provides a robust solution for underdetermined wideband DOA estimation using coprime arrays.
• The fixed-point update approach in SBL effectively handles dictionary mismatch.
• SBL offers superior performance in both detection and estimation accuracy compared to existing sparse methods.