Efficient rank-recovery-based coherent source localization framework for non-uniform FDA

Abstract

This article addresses the problem of resolving coherent sources in non-uniform frequency diverse arrays (FDAs), where existing decoherence methods fail due to the unique geometric irregularity. We propose a novel covariance matrix reconstruction framework that enables high-resolution joint estimation of range and angle. The key innovation lies in a dual-structure recovery mechanism: First, a binary mask matrix is designed using FDA-specific space–frequency difference constraints to restore the degraded sample covariance’s Hermitian-Toeplitz structure. Atomic norm minimization is then integrated to achieve super-resolution parameter estimates, with the alternating direction method of multipliers enabling computationally efficient optimization. Theoretical analysis establishes performance bounds for covariance matrix reconstruction, while extensive simulations demonstrate the proposed method’s superior estimation accuracy over conventional subspace-based approaches in coherent scenarios, while maintaining low computational complexity.