Abstract:Direction of arrival (DoA) estimation is a fundamental task in array processing. A popular family of DoA estimation algorithms are subspace methods, which operate by dividing the measurements into distinct signal and noise subspaces. Subspace methods, such as Multiple Signal Classification (MUSIC) and Root-MUSIC, rely on several restrictive assumptions, including narrowband non-coherent sources and fully calibrated arrays, and their performance is considerably degraded when these do not hold. In this work we propose SubspaceNet; a data-driven DoA estimator which learns how to divide the observations into distinguishable subspaces. This is achieved by utilizing a dedicated deep neural network to learn the empirical autocorrelation of the input, by training it as part of the Root-MUSIC method, leveraging the inherent differentiability of this specific DoA estimator, while removing the need to provide a ground-truth decomposable autocorrelation matrix. Once trained, the resulting SubspaceNet serves as a universal surrogate covariance estimator that can be applied in combination with any subspace-based DoA estimation method, allowing its successful application in challenging setups. SubspaceNet is shown to enable various DoA estimation algorithms to cope with coherent sources, wideband signals, low SNR, array mismatches, and limited snapshots, while preserving the interpretability and the suitability of classic subspace methods.

Novel High-Resolution DOA Estimation Using Subspace Projection Method

Efficient Algorithm for 2-D DOA Estimation Based on Noise Subspace Mapping

Low-complexity DOA Estimation via Subspace Rotation Technique

Low-Complexity DOA Estimation Based on Compressed MUSIC and Its Performance Analysis

A High Resolution Algorithm for Nonplanar Array with Arbitrary Geometry Without Source Number Estimation

Improved Subspace Direction-of-arrival Estimation in Unknown Nonuniform Noise Fields

Direction of Arrival Estimation Method Based on the Angular Signal Subspace

Method to Estimate and Promote SNR of Covariance Matrix in Array Signal Processing

Analysis of Subspace Projection DOA Algorithms and Synthetic Spectrum

Modified MUSIC Approach with Weighted Pseudo-Noise Subspace Projection

An oblique projection filtering based DOA estimation algorithm without a priori knowledge

New high-resolution DOA estimation algorithm

Signal-subspace high-resolution direction finding based on orthogonal projections

Modified MUSIC Algorithm for DOA Estimation with Nyström Approximation

High resolution MUSIC algorithm reconstructing covariance matrix in low SNR

Direction of Arrival Estimation of Wideband Signals with Sensor Arrays in Spatial Non-stationary Noise

2-D DOA estimation method based on dimension descended noise subspace

SubspaceNet: Deep Learning-Aided Subspace Methods for DoA Estimation

Spatial Aliasing for Efficient Direction-of-arrival Estimation Based on Steering Vector Reconstruction

Reduced-complexity Direction of Arrival Estimation with Centro-Symmetrical Arrays and Its Performance Analysis

Subspace-Based Direction-Of-Arrival Estimation Using Centro-Symmetrical Arrays