Phase–Amplitude Fusion Network for Underwater Environment Noise Suppression in Fiber-Optic Hydrophones
- 1 May 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Letters
- Vol. 10 (5) , 1-4
Abstract
The presence of environment noise severely limits the detection accuracy of fiber-optic hydrophones (FOHs). Mask-based learning methods have attracted significant attention for environment noise suppression, but most existing approaches focus on amplitude information only. In this work, we investigate using a phase–amplitude fusion network (PHASEN) for underwater environment noise suppression. To validate the effectiveness of simultaneously exploiting phase and amplitude information, an ablation study is conducted. Furthermore, the performance of PHASEN is compared with conventional denoising methods using noise data physically collected from a custom-built Michelson interferometer-based FOH platform, demonstrating its superior denoising capability. Finally, the improvement in underwater acoustic target recognition from using PHASEN is evaluated. Results in this paper highlight the importance of phase information and demonstrate the potential of PHASEN for improving the robustness and deployment capability of practical underwater sensing.Keywords
This publication has 14 references indexed in Scilit:
- Bearing-Only Tracking Considering Dynamics of a Towed Sensor-ArrayIEEE Sensors Letters, 2024
- Multimode Microstructured Optical Fiber HydrophoneIEEE Sensors Journal, 2024
- Ultrasound signal processing based on joint GWO-VMD wavelet threshold functionsMeasurement, 2024
- Underwater acoustic signal denoising model based on secondary variational mode decompositionDefence Technology, 2023
- A review of the state of research on wave-current interaction in nearshore areasOcean Engineering, 2022
- DeepShip: An underwater acoustic benchmark dataset and a separable convolution based autoencoder for classificationExpert Systems with Applications, 2021
- Physical Sources of High‐Frequency Seismic Noise on Cascadia Initiative Ocean Bottom SeismometersGeochemistry, Geophysics, Geosystems, 2020
- Supervised Speech Separation Based on Deep Learning: An OverviewIEEE/ACM Transactions on Audio, Speech, and Language Processing, 2018
- Deep Residual Learning for Image Recognition2016
- Speech Enhancement with LSTM Recurrent Neural Networks and its Application to Noise-Robust ASR2015