.png){kind=icon}
Return
Dynamic weighted linear prediction-based on DOA estimation using acoustic vector sensor array on moving platforms
.png){kind=icon}
PRE
AI
DOI：10.1016/j.dsp.2026.106464.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
To overcome the challenge of limited direction of arrival (DOA) estimation performance on small-scale underwater moving platforms, a dynamic weighted linear prediction (DWLP) method for acoustic vector sensor array (AVSA) is proposed in this paper. First, using the platform’s motion, an equivalent virtual AVSA is constructed via passive synthetic aperture technology. To fill the resulting AVSA gaps, spline interpolation is applied to recover missing data, thereby reconstructing the non-uniform array into a uniform linear array. Subsequently, based on the linear correlation of the uniform linear array data, a linear prediction (LP) model is established to simultaneously generate virtual acoustic vector sensor (AVS) at both ends of the AVSA. In the prediction process, a dynamic weighting matrix and an error screening mechanism are introduced to eliminate abnormal predicted values, effectively improving the accuracy of the virtual AVS data. To mitigate steering vector mismatch caused by prediction errors, the covariance matrix is reconstructed through eigenanalysis, and the steering vector deviation is corrected using Taylor series expansion. On this basis, an iterative loop is further employed to refine the DOA estimation. Simulation results demonstrate that, compared with existing array aperture extension methods, the proposed DWLP method achieves higher DOA estimation accuracy and superior noise suppression performance under the considered small-scale moving platform configuration.
Keywords:
Acoustic vector sensor array (AVSA)
Array interpolation
Dynamically weighted linear prediction (DWLP)
Direction of arrival (DOA) estimation
Moving platforms
Journal
D
IF:
3
Papers:
693
Citations:
0
Organization
View more organizations
Cited Papers
No cited papers available