Revisiting Deep Augmented Music Algorithm
In this paper, we revisit the hybrid DA-MUSIC model, which combines the classical MUSIC algorithm with deep neural networks to improve the estimation of direction of arrival (DoA). By incorporating Hermitian and positivity constraints into the model covariance matrix, we achieve a performance enhancement. Additionally, we introduce a novel data-driven model that employs recurrent neural networks and multilayer perceptrons, demonstrating superior performance compared to DA-MUSIC across various scenarios.
Authors
- Abdeldjalil Aïssa El Bey (ORCID: https://orcid.org/0000-0002-6267-3118)
- H. T. Huynh (ORCID: https://orcid.org/0000-0002-1182-1731)
- Thierry Chonavel (ORCID: https://orcid.org/0000-0003-3406-0426)
Institutions
- Centre National de la Recherche Scientifique (FR)
- Université de Bretagne Occidentale (FR)
- Laboratoire des Sciences et Techniques de l’Information de la Communication et de la Connaissance (FR)
- IMT Atlantique (FR)
Publication Details
- Published
- 2025-09-08
- DOI
- https://doi.org/10.23919/eusipco63237.2025.11226743
- Primary Topic
- Direction-of-Arrival Estimation Techniques
- Type
- article
- Field-Weighted Citation Impact
- 0.00