Circulant ADMM-Net for Fast High-resolution DoA Estimation

2025/02/26 by Youval Klioui, Klioui, Youval · 1 citation
#eess.SP

paper · pdf · doi:10.48550/arxiv.2502.19076

Abstract

This paper introduces CADMM-Net and CHADMM-Net, two deep neural networks for direction of arrival estimation within the least-absolute shrinkage and selection operator (LASSO) framework. These two networks are based on a structured deep unfolding of the alternating direction method of multipliers (ADMM) algorithm through the use of circulant as well as Hermitian-circulant matrices. Along with a computational complexity of O(Nlog(N)) per layer for the inference, where N is the length of the dictionary A, they additionally exhibit a memory footprint of N and approximately half of N for CADMMNet and CHADMM-Net, respectively, compared with N² for ADMM-Net. Furthermore, these structured networks exhibit a competitive performance against ADMM-Net, LISTA, TLISTA, and THLISTA with respect to the detection rate, the angular root-mean square error, and the normalized mean squared error.

Citations

• Maximum likelihood methods for direction-of-arrival estimation
• ESPRIT-estimation of signal parameters via rotational invariance techniques
• Forward/backward spatial smoothing techniques for coherent signal identification

Cited by

• Fast Single-Snapshot Harmonic Recovery with 2D Sparse Arrays using BCCB Matrices

Related

• Toeplitz-Hermitian ADMM-Net for DoA Estimation
• AA-DLADMM: An Accelerated ADMM-based Framework for Training Deep Neural Networks
• Direction-of-arrival estimation with conventional co-prime arrays using deep learning-based probablistic Bayesian neural networks
• BADM: Batch ADMM for Deep Learning
• A Convergent ADMM Framework for Efficient Neural Network Training