This article investigates the problem of direction-of-arrival estimation based on sparse recovery (SR). Unlike convex-relaxation-based SR methods, the reweighted atomic-norm minimization (RAM) introduces a nonconvex sparsity metric that enhances sparsity and resolution. However, it solely embeds the log-det penalty and exhibits parameter redundancy. In this study, we propose a generalized reweighted $\ell _{2,1}$ minimization approach named gridless reweighted row-norm minimization (GRRM). A gridless formulation for the weighted $\ell _{2,1}$ minimization is first derived, based on which a reweighting framework is proposed with details regarding its implementation. Besides, this framework is also extended to accommodate nonuniform arrays. Compared to RAM, GRRM is capable of effectively incorporating widely used nonconvex functions while requiring a reduced number of parameters for estimation. Both simulation and field experiments are conducted to demonstrate that GRRM performs comparably to RAM with reduced computational complexity and exhibits superior performance compared to the subspace-based and convex-relaxation-based SR methods.
Generalized Gridless Formulation of Reweighted $\ell _{2,1}$ Minimization for DoA Estimation
IEEE Transactions on Aerospace and Electronic Systems ; 61 , 2 ; 2295-2308
01.04.2025
4712472 byte
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
British Library Conference Proceedings | 2005
|Development of the linear gridless ion thruster
AIAA | 2001
|Reweighted sparse subspace clustering
British Library Online Contents | 2015
|