A modified vision transformer based direction of arrival estimation using single vector sensor

Seo-jun Choi; Jae-geun Lee; Ki-man Kim; Ji-won Jung

doi:10.7776/ASK.2026.45.5.536

All Issue

2026 Vol.45, Issue 5 Preview Page Next Page

A modified vision transformer based direction of arrival estimation using single vector sensor 단일 벡터 센서를 사용한 변형된 비전 트랜스포머 기반 도래각 추정

30 September 2026. pp. 536-541

PDF XML

Abstract

The acoustic vector sensor is one of the specialized sensors capable of estimating the direction of a sound source on its own. This paper proposes a method for estimating the direction of a sound source based on deep learning in an environment utilizing only a single vector sensor. The proposed method reduces the Direction of Arrival (DOA) estimation error by inputting a covariance matrix and an intensity based histogram instead of images in a Vision Transformer (ViT)-based method, which is one of the existing deep learning methods. In the simulations performed to verify the performance, it was assumed that the vector sensor has two dimensional particle velocity output with acoustic pressure. The simulation results show that the proposed method exhibits superior performance compared to existing representative methods, and particularly demonstrates improved accuracy in estimating the direction of arrival in environments with a low Signal to Noise Ratio (SNR). In simulations, it was confirmed that the proposed method reduces the Root Mean Squared Error (RMSE) by approximately 30 % compared to the existing method in estimating the direction of arrival when the Signal to Noise Ratio is –10 dB.

Keywords

Vector sensor, Direction of Arrival (DOA) estimation

Deep learning

Vision Transformer (ViT)

Covariance matrix

Root Mean Squared Error (RMSE)

음향 벡터 센서는 그 하나만으로도 음원의 방향을 추정할 수 있는 특수한 센서 가운데 하나이다. 이 논문에서는 한 개의 벡터 센서만을 이용하는 환경에서 심층 학습에 기반하여 음원의 방향을 추정하는 방법을 제안한다. 제안하는 방법은 기존의 심층 학습 가운데 하나인 비전 트랜스포머(Vision Transformer, ViT) 기반의 방법에서 이미지 대신 공분산 행렬과 인텐시티 기반 히스토그램을 입력함으로써 도래각(Direction of Arrival, DOA) 추정 오차가 감소하도록 하였다. 성능을 검증하기 위해 수행된 모의실험에서는 벡터 센서가 음향 압력과 함께 2차원 입자 속도 출력을 갖는다고 가정하였다. 모의실험 결과는 제안한 방법이 기존의 대표적인 방법들보다 우수한 성능을 보이며, 특히 낮은 신호대잡음비(Signal to Noise Ratio, SNR)를 갖는 환경에서 도래 방향 추정의 정확도가 향상됨을 보여준다. 모의실험에서 제안한 방법은 신호대잡음비가 –10 dB인 경우 도래각 추정에 있어서 기존 방식 대비 평균 제곱근 오차(Root Mean Squared Error, RMSE)가 약 30 % 감소하는 것을 확인하였다.

키워드

벡터 센서

도래각(Direction of Arrival, DOA) 추정

심층 학습, 비전 트랜스포머(Vision Transformer, ViT)

공분산 행렬

평균 제곱근 오차(Root Mean Squared Error, RMSE)

References

Y. S. Lim and Y. R. Roh, “Incidence angle estimation by the Tonpilz type underwater acoustic vector sensor with a quadrupole structure” (in Korean), J. Acoust. Soc. Kr. 31, 569-579 (2012).

10.7776/ASK.2012.31.8.569

A. Nehorai and E. Paldi, “Acoustic vector-sensor array processing,” IEEE Trans. Signal Process. 42, 2481-2491 (1994).

10.1109/78.317869

M. Hawkes and A. Nehorai, “Acoustic vector-sensor beamforming and Capon direction estimation,” IEEE Trans. Signal Process. 46, 2291-2304 (1998).

10.1109/78.709509

H.-W. Chen and J.-W. Zhao, “Wideband MVDR beamforming for acoustic vector sensor linear array,” IEE Proc. Radar Sonar Navig. 151, 158-162 (2004).

10.1049/ip-rsn:20040651

Y. Xie and B. Wang, “Direction-of-arrival estimation method based on neural network with temporal structure for underwater acoustic vector sensor array,” Sensors, 23, 1-15 (2023).

10.3390/s2310491937430832PMC10222060

S. Whitaker, A. Barnard, G. D. Anderson, and T. C. Havens, “Through-ice acoustic source tracking using vision transformers with ordinal classification,” Sensors, 22, 1-20 (2022).

10.3390/s2213470335808200PMC9269127

H. Cao, W. Wang, L. Su, H. Ni, P. Gerstoft, Q. Ren, and L. Ma, “Deep transfer learning for underwater direction of arrival using one vector sensor,” J. Acoust. Soc. Am. 149, 1699-1711 (2021).

10.1121/10.0003645

A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, “An image is worth 16 × 16 words: Transformers for image recognition at scale,” Proc. ICLR, 1-22 (2021).

A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, L. Kaiser, and I. Polosukhin, “Attention is all you need,” Proc. 31st NIPS, 5998-6008 (2017).

Information

Publisher :The Acoustical Society of Korea
Publisher(Ko) :한국음향학회
Journal Title :The Journal of the Acoustical Society of Korea
Journal Title(Ko) :한국음향학회지
Volume : 45
No :5
Pages :536-541
Received Date : 2026-07-30
Accepted Date : 2026-08-18
DOI :https://doi.org/10.7776/ASK.2026.45.5.536

The Journal of the Acoustical Society of KoreaISSN:1225-4428(Print) 2287-3775(Online)한국음향학회

All Issue