An Efficient Direct Position Determination Method for Multiple Strictly Noncircular Sources

Area of Science:

• Signal Processing
• Array Signal Processing
• Electromagnetics

Background:

• Conventional localization methods use a two-step approach: parameter extraction then position estimation.
• Direct Position Determination (DPD) offers higher accuracy by localizing sources directly from sensor data, avoiding intermediate parameters and data association.
• Existing Subspace Data Fusion (SDF)-based DPD methods are computationally efficient but do not leverage signal-specific properties.

Purpose of the Study:

• To propose an improved SDF-based DPD algorithm tailored for strictly noncircular sources.
• To enhance the localization accuracy and computational efficiency of DPD methods.
• To exploit the unique properties of noncircular signals for improved source localization.

Main Methods:

• Derived the frequency-domain properties of strictly noncircular signals.
• Extended frequency-domain observation vectors by exploiting noncircularity to create extended noise subspaces.
• Fused extended noise subspaces across all arrays and frequencies, then applied a unitary transformation to formulate a source location cost function based on eigenvalues.
• Developed a Newton-type iterative method using matrix Eigen-perturbation theory to efficiently solve the nonlinear cost function, avoiding grid search.

Main Results:

• The proposed DPD algorithm significantly reduces running time compared to existing methods.
• Achieved superior localization performance for both near-field and far-field noncircular sources.
• Demonstrated the effectiveness of exploiting noncircular signal properties within the SDF-DPD framework.

Conclusions:

• The improved SDF-based DPD algorithm offers a computationally efficient and accurate solution for localizing multiple strictly noncircular sources.
• The integration of signal properties and advanced iterative methods enhances DPD performance.
• This approach provides a robust method for source localization with widely separated arrays.