Residual network based Doppler shift frequency estimation using spectrogram images

Ji-hyun Lee; Jae-keun Lee; Ki-man Kim

doi:10.7776/ASK.2026.45.4.435

All Issue

2026 Vol.45, Issue 4 Preview Page

Residual network based Doppler shift frequency estimation using spectrogram images 스펙트로그램 이미지를 사용한 잔차 네트워크 기반의 도플러 천이 주파수 추정

31 July 2026. pp. 435-444

PDF XML

Abstract

In underwater acoustic communications, maintaining high communication performance in dynamic environments where transceivers move is crucial. This paper proposes a Residual Network (ResNet) based regression method that utilizes spectrogram images of the preamble to estimate the Doppler shift frequency, a key factor determining the performance of underwater acoustic communications. Unlike the previous Convolutional Neural Network (CNN) based classification methods, the proposed approach enables high resolution estimation by directly estimating the Doppler shift frequency as a continuous value. The ResNet architecture employs skip connections that allows stable learning of subtle differences caused by Doppler effect. A preamble structure composed of concatenated down and up sweep Linear Frequency Modulation (LFM) signals are employed to effectively capture time and frequency variations caused by Doppler effects. Simulation results demonstrate that the proposed method outperforms CNN based approach adapted for regression. Specifically, the proposed method reduces the average Doppler shift frequency estimation error by approximately 2 Hz compared to the CNN baseline at an Signal to Noise Ratio (SNR) of 20 dB in Bellhop simulations.

Keywords

Doppler shift frequency

Deep learning

Residual Network (ResNet)

Convolutional Neural Network (CNN)

Linear Frequency Modulation (LFM)

수중음향통신에 있어서 송수신기가 이동하는 동적 환경에서 높은 통신 성능을 유지하는 것이 필요하다. 이 논문은 수중음향통신의 성능을 결정하는 주요 요인 가운데 하나인 도플러 천이 주파수를 추정하기 위해 프리앰블 구간의 스펙트로그램 이미지를 이용하여 잔차 네트워크(Residual Network, ResNet) 기반 회귀 추정 기법을 제안한다. 기존 합성곱 신경망(Convolutional Neural Network, CNN) 기반의 분류 방식과는 다르게 제안하는 기법은 도플러 천이 주파수를 연속적인 값으로써 직접 추정함으로써 높은 해상도의 추정이 가능하다. 잔차 네트워크는 잔차 연결 구조를 통해 도플러 효과에 의한 미세한 차이를 안정적으로 학습을 가능하게 한다. 또한 하향 및 상향 선형 주파수 변조(Linear Frequency Modulation, LFM) 신호를 연접한 프리앰블 구조를 활용하여 도플러 효과에 따른 시간 및 주파수 변화를 효과적으로 반영하였다. 모의실험 결과를 통해 제안된 방법이 기존의 분류 합성곱 신경망을 회귀 기반으로 변형한 방법보다 향상된 성능을 달성함을 보여준다. 제안한 방법은 Bellhop 시뮬레이션에서 신호대잡음비(Signal to Noise Ratio, SNR)가 20 dB인 경우 기존 방식 대비 약 2 Hz의 평균 도플러 천이 주파수 추정 오차가 감소하는 것을 확인하였다.

키워드

도플러 천이 주파수

심층 학습

잔차 네트워크

합성곱 신경망

선형 주파수 변조

References

R. Diamant, A. Feuer, and L. Lampe, “Choosing the right signal: Doppler shift estimation for underwater acoustic signals,” Proc. Int. Conf. Underwater Netw. Syst. 1-8 (2012).

10.1145/2398936.2398971

A. M. Bassam and M. L. Seto, “Derivation of a Low-Complexity-Pilot-Aided Doppler estimator for underwater acoustic applications,” J. Acoust. Soc. Am.157, 3705-3717 (2025).

10.1121/10.0036691

A. D. Waite, Sonar for Practising Engineers, 3rd Ed (John Willey & Sons, New Jersey, 2002), pp. 169-170.

B. S. Sharif, J. Neasham, O. R. Hilton, and A. E. Adams, “A computationally efficient Doppler compensation system for underwater acoustic communications,” IEEE J. Oceanic Eng. 25, 52-61 (2000).

10.1109/48.820736

D. Sun, K. Hou, and T. Teng, “Underwater weak moving target detection method based on wideband multi-pulse coherent integration,” Appl. Acoust. 212, 109616 (2023).

10.1016/j.apacoust.2023.109616

S. Kojima, K. Maruta, Y. Feng, C. J. Ahn, and V. Tarokh, “CNN-based joint SNR and Doppler shift classification using spectrogram images for adaptive modulation and coding,” IEEE Trans. Commun. 69, 5152-5165 (2021).

10.1109/TCOMM.2021.3077565

K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” Proc. IEEE Conf. Comput. Vis. Pattern Recognit. 770-778 (2016).

10.1109/CVPR.2016.90

A. Krizhevsky, I. Sutskever, and G. E. Hinton, “ImageNet classification with deep convolutional neural networks,” Proc. Adv. Neural Inf. Process. Syst. 1097-1105 (2012).

D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” Proc. Int. Conf. Learning Representations, 1-15 (2015).

Y. Gal and Z. Ghahramani, “Dropout as a Bayesian approximation: Representing model uncertainty in deep learning,” Proc. Int. Conf. Mach. Learn. 1050-1059 (2016).

P. Kotipalli, A. S. M. Mohanraju, and P. Vardhanapu, “Frame boundary detection and deep learning-based Doppler shift estimation for FBMC/OQAM communication system in underwater acoustic channels,” IEEE Access, 10, 17590-17608 (2022).

10.1109/ACCESS.2022.3148410

Y. Zhang, Y. Wang, Y. Liu, L. Shi, and Y. Zang, “A deep learning receiver for underwater acoustic OTFS communications with Doppler squint effect,” IEEE Wireless Commun. Lett. 14, 1179-1183 (2025).

10.1109/LWC.2025.3538017

S. H. Oh, H. S. Kim, J. S. Kim, J. H. Cho, J. H. Chung, and H. C. Song, “Performance analysis of underwater acoustic communication systems using underwater channel simulation tool” (in Korean), J. Acoust. Soc. Kr. 31, 373-383 (2012).

10.7776/ASK.2012.31.6.373

D. P. Williams, “On the use of tiny convolutional neural networks for human-expert-level classification performance in sonar imagery,” IEEE J. Oceanic Eng. 46, 236-260 (2020).

10.1109/JOE.2019.2963041

M. Stojanovic, “Recent advances in high-speed underwater acoustic communications,” IEEE J. Oceanic Eng. 21, 125-136 (1996).

10.1109/48.486787

M. Siderius and M. B. Porter, “Modeling broadband ocean acoustic transmissions with time-varying sea surfaces,” J. Acoust. Soc. Am. 124, 137-150 (2008).

10.1121/1.2920959

B. Shi, X. Bai, and C. Yao, “An end-to-end trainable neural network for image-based sequence recognition and its application to scene text recognition,” IEEE Trans. Pattern Anal. Mach. Intell. 39, 2298-2304 (2017).

10.1109/TPAMI.2016.2646371

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 :4
Pages :435-444
Received Date : 2026-02-25
Revised Date : 2026-04-01
Accepted Date : 2026-04-16
DOI :https://doi.org/10.7776/ASK.2026.45.4.435

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

All Issue