A linearly interpolated DOA estimation algorithm based on Variational Bayesian Inference

• Published in: Applied acoustics Vol. 240; p. 110968
• Main Authors: Cui, Lin, Cui, Yingkai, Liu, Jialei
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 05.12.2025
• Subjects: Direction-of-arrival (DOA) estimation; Linear Interpolation; Sparse Bayesian Learning (SBL); Temporal Correlation structure; Variational Bayesian Inference; Variational Bayesian Inference; Direction-of-arrival (DOA) estimation; Temporal Correlation structure; Linear Interpolation; Sparse Bayesian Learning (SBL)
• ISSN: 0003-682X
• DOI: 10.1016/j.apacoust.2025.110968

•Acoustic source direction of arrival estimation based on Variational Bayesian Inference.•A grid interpolation algorithm to perform off-grid direction of arrival estimation.•This algorithm still performs well under the condition of a coarse grid. Acoustic source localization constitutes a pivotal research subject within the domain of Direction-of-Arrival (DOA) estimation. In particular, DOA estimation algorithms based on Sparse Bayesian Learning (SBL) tend to experience severe performance degradation when the chosen grid intervals are large. To solve this problem, this paper proposes a DOA estimation algorithm named Linear Interpolation Temporal Correlation − Variational Bayesian Inference (LITC-VBI) based on Variational Bayesian Inference (VBI). Within the framework of the VBI, a novel signal model is constructed by exploiting the inherent temporal correlation of the incident acoustic signal. In addition, by combining the linear interpolation method, a new off-grid model is constructed. This model improves the accuracy and stability of DOA estimation in the off-grid scenario. Simulation results demonstrate that the DOA estimation performance is significantly improved and that it outperforms several existing SBL-based DOA estimation methods.