Robust AOA-Based Target Localization for Uniformly Distributed Noise via ℓ-ℓ₁ Optimization

Area of Science:

• Signal Processing
• Estimation Theory
• Optimization

Background:

• Robust target localization is crucial in various applications.
• Uniformly distributed noise, a mixture of Laplacian and uniform distributions, poses significant challenges.
• Existing methods may lack robustness or sensitivity to noise parameters.

Purpose of the Study:

• To develop novel algorithms for robust angle of arrival (AOA) target localization.
• To address challenges posed by mixed uniform and Laplacian noise distributions.
• To improve localization accuracy and reduce sensitivity to noise parameters.

Main Methods:

• A novel localization model using ℓp-norm for measurement error and ℓ1-norm for regularization.
• Development of a proximal operator alternating direction method of multipliers (POADMM) for convex problems (1≤p<2).
• Proposal of a smoothed POADMM (SPOADMM) estimator for nonconvex and nonsmooth problems (0≤p<1).

Main Results:

• The proposed POADMM and SPOADMM algorithms effectively solve the derived optimization problems.
• Numerical simulations show superior performance compared to state-of-the-art robust algorithms.
• The methods exhibit enhanced robustness against outliers and reduced sensitivity to measurement noise variance.

Conclusions:

• The developed POADMM and SPOADMM algorithms offer effective solutions for robust AOA target localization.
• These methods provide significant improvements in performance under challenging noise conditions.
• The proposed approach is less sensitive to parameter selection, enhancing practical applicability.