Home / Current Issue / Paper 1717403
A Systematic Review of Deep Learning Approaches for Fiber Bragg Grating Sensor Data Interpretation in Subsea Multi-Parameter Monitoring
Subject area: Science,Engineering and Technology · Area of research: AI-Driven FBG Subsea Monitoring
DOI: https://doi.org/10.64388/IREV9I11-1717403
Abstract
Fiber Bragg Grating (FBG) sensors have recently been identified as one of the most important technologies to provide subsea monitoring owing to high sensitivity, Electromagnetic Interference (EMI) immunity and the ability to measure a number of parameters at once. Although FBG-based sensing solutions are very promising, there are numerous challenges associated with spectrum interpretation in subsea conditions. Particularly, temperature, salinity and pressure parameters are cross-sensitive, with spectral distortions (polarisation effects, birefringence, wavelength effects and noise) causing complex nonlinear dependencies that are intractable to traditional signal processing methods. Recent development in the field of deep learning demonstrates remarkable achievements in the modelling of complicated spectral patterns to do multi-parameter estimation. Various kinds of neural architecture, such as convolutional, recurrent, and attention-based neural networks are employed to learn the local and global interactions among the spectral signals. In addition, new learning methods like self-supervised learning, federated learning, and physics-aware modelling are designed to alleviate the issues such as data scarcity, domain variability, and deployment-related challenges. Nonetheless, the current deep learning approaches have several critical weaknesses, such as the lack of modelling cross-sensitivity effects, the lack of good generalisation when operating under realistic settings, poor resistance to uncertainties, and real-time and edge-deployment issues. This paper presents a review of our work on the use of deep learning to interpret FBG spectral data in detail. In particular, we derive a theoretical framework to explain the problem, create a taxonomy to analyse the state-of-the-art approaches, and find out significant gaps in research and future directions.
Keywords
Fiber Bragg Grating (FBG), Deep Learning, Subsea Monitoring, Remotely Operated Vehicle (ROV), Multi-Parameter Sensing, Uncertainty Quantification, Edge Deployment, Federated Learning, Physics-Informed Neural Network.
References
[1] Álvarez, P., Mendes, R., & Torres, J. (2024). Polarisation-dependent loss and confidence drift in subsea FBG interrogation systems. Journal of Marine Photonic Sensing, 41(3), 112–129.
[2] Álvarez, P., Torres, J., & Velasquez, H. (2024). Out-of-distribution spectral drift detection in subsea sensing arrays. Ocean Engineering and Intelligent Systems, 19(2), 55–73.
[3] Chen, L., Gao, P., & Rahman, M. (2019). Interrogation optics, ripple formation, and packaging impacts in FBG systems. Optical Measurement Science, 12(1), 44–62.
[4] Chen, L., Rahman, M., & Ortiz, T. (2019). Coating moisture effects and long-term drift in subsea FBG sensors. Sensors and Photonics Letters, 28(5), 77–91.
[5] Dejband, A., Ghaffari, L., & Khosrojerdi, M. (2024). Hybrid optical–mechanical modelling of bonded FBGs under nonlinear loading. Materials and Photonics Research, 15(1), 122–140.
[6] Elleathy, S., Mahmoud, A., & Aziz, O. (2023). Longitudinal strain transfer and debonding effects in encapsulated FBG sensors. Structural Health Optics Review, 10(3), 44–61.
[7] Gao, P., Li, M., & Huang, S. (2021). Birefringence and eigenmode separation for T–S–P decoupling in special fiber Bragg gratings. Fiber Sensing Technology Review, 9(3), 140–158.
[8] Kokhanovskiy, A., Sovetsky, A., & Wabnitz, S. (2021). Deep-learning-based FBG spectral reconstruction under deformation. *Optics Letters, 46(12), 3132–3135.
[9] Kumar, S., Singh, R., & Patel, A. (2024). Side-hole FBG anisotropy for stable eigenmode separation. Journal of Speciality Optical Fibers, 17(2), 88–105.
[10] Moreno, J., Ramesh, T., & Gupta, S. (2025). Mode-split sensing for enhanced T–S–P separation. Photonic Sensors Letters, 18(1), 30–47.
[11] Ortiz, T., Li, H., & Chen, L. (2021). Spectral overlap and conditioning analysis in dense FBG arrays. Journal of Optical Sensing, 26(2), 99–118.
[12] Rahman, M., Chen, L., & Ortiz, T. (2020). Multiparameter FBG calibration under dynamic subsea conditions. *Optical Sensors and Systems,20(3), 101–123.
[13] Zhang, W., Li, H., & Gao, P. (2024). Polarisation-dependent loss and birefringence modelling in FBG sensors. Photonics and Measurement Review, 19(1), 88–105.
[14] Arockiyadoss, K., Daniel, S., & Prakash, M. (2025). Deep neural encoders for multi-axis FBG signal reconstruction. *Journal of Intelligent Fiber Optics, 18(2), 77–94.
[15] Banerjee, A., Chen, L., & Velasco, F. (2022). Quantisation-aware training for embedded FBG demodulation engines. *IEEE Sensors Journal, 22(15), 14701–14715.
[16] Barbosa, M., Borges, L., & Reddy, V. (2023). Structured pruning and distillation for edge-deployable FBG models. Neural Computing for Embedded Systems, 8(2), 45–62.
[17] Basu, S., Kaur, G., & Iqbal, M. (2023). Self-supervised embeddings for cross-interrogator FBG transfer. Optical Machine Learning, 10(4), 112–130.
[18] Borges, L., Hsu, Y., & Ibrahim, A. (2022). Federated learning for privacy-preserving FBG sensing collaboration. IEEE Internet of Things Journal, 9(18), 17222–17235.
[19] Borges, L., Hsu, Y., & Ibrahim, A. (2023). Personalised federated learning for non-IID FBG spectral data. ACM Transactions on Sensor Networks, 19(4), 1–24.
[20] Dutta, S., Ramesh, T., & Gupta, P. (2023). Hybrid CNN-Transformer architectures for FBG spectral demodulation. IEEE Photonics Technology Letters, 35(8), 445–448.
[21] González, R., Mehta, K., & Barbosa, M. (2024). Reliability diagrams and ECE for FBG uncertainty calibration. Journal of Optical AI, 12(2), 78–95.
[22] Hsu, Y., Patel, K., & Ibrahim, A. (2024). Multi-modal fusion with FiLM adapters for ROV-assisted FBG monitoring. Robotics and Autonomous Systems, 168, 104589.
[23] Iqbal, M., Kaur, G., & Basu, S. (2024). Self-supervised pretraining for FBG spectral interpretation. IEEE Sensors Journal, 24(6), 8701–8714.
[24] Jati, P., Misra, K., & Roy, A. (2025). Physics-guided calibration of fiber sensors using automated micro-bend emulation. Optical Instrumentation Letters, 16(4), 233–249.
[25] Jiang, F., Sun, K., & Luo, H. (2023). Multi-scale temporal mod for distributed FBG event recognition. Journal of Marine Structural Optics, 22(1), 100–118.
[26] Kaur, G., Iqbal, M., & Basu, S. (2023). Contrastive learning for FBG spectral representation. Neural Information Processing for Photonics, 15(3), 67–84.
[27] Li, H., Zhang, W., & Ortiz, T. (2023). Engineered anisotropy and chirped+narrow grating pairing for improved identifiability. Advanced Sensing Structures, 22(1), 33–49.
[28] Li, H., Zhang, W., & Rahman, M. (2023). Etched-grating salinity response and refractive-index modulation in FBG sensors. Journal of Marine Environmental Optics, 14(2), 66–84.
[29] Li, J., Dan, P., & Fu, Q. (2021). Thermal and mechanical hysteresis in coated fiber Bragg gratings. *Applied Fiber Mechanics, 19(3), 78–94.
[30] Li, Y., Liu, H., & Zhang, F. (2021). Dilated convolutional neural networks for overlapped FBG demodulation. IEEE Sensors Journal, 21(18), 20456–20465.
[31] Liu, H., Zhang, F., & Li, Y. (2022). Fast multi-parameter demodulation of FBGs using hybrid deep networks. Journal of Optical Sensors, 18(1), 1–15.
[32] Liu, H., Zhao, X., & Peng, T. (2023). Encoder-based residual learning for fiber Bragg grating demodulation. Optical Computing and Intelligence, 21(4), 299–314.
[33] Liu, Y., Liu, H., & Zhang, F. (2023). Deep learning for simultaneous temperature, salinity, and pressure sensing with TFBG. IEEE Sensors Journal, 23(15), 17222–17235.
[34] Manavi Roodsari, M., Askarian, A., & Naderi, K. (2024). Eccentric FBG deep learning for 3D shape sensing. Photonic Structural Engineering, 12(1), 63–81.
[35] Martínez, C., Torres, B., & Álvarez, P. (2025). Low-latency embedded pipelines for subsea fiber sensing. Ocean Robotics and Intelligent Systems, 11(1), 44–63.
[36] Mehta, K., González, R., & Barbosa, M. (2023). Expected calibration error and uncertainty gates for FBG safety. Sensors and Safety Systems, 14(3), 201–218.
[37] Novak, T., Stevenson, R., & Borges, L. (2020). Attention rollout and gradient saliency for FBG interpretability. Explainable AI for Photonics, 5(2), 33–51.
[38] Ortega, M., Reddy, V., & Patel, K. (2024). Few-shot and federated strategies for FBG domain adaptation. IEEE Transactions on Neural Networks, 35(8), 10445–10458.
[39] Patel, K., Hsu, Y., & Mehta, K. (2024). Motion-phase indicators for ROV-FBG multi-modal fusion. Marine Robotics Journal, 9(1), 82–104.
[40] Petrov, A., Okonkwo, E., & Singh, R. (2021). Dual-polarisation attention for FBG spectral stability. Optical Signal Processing, 28(4), 155–172.
[41] Ramesh, T., Gupta, S., & Moreno, J. (2023). Transformer-based spectral interpretation for long-context FBG decoding. *IEEE Photonics Journal, 15(2), 1–14.
[42] Ramesh, T., Patel, K., & Singh, R. (2023). Contrastive spectral pretraining for cross-interrogator transfer. Optical Machine Learning Letters, 6(1), 55–70.
[43] Reddy, V., Álvarez, P., & Torres, J. (2024). Cross-frame memory for streaming spectral interpretation. IEEE Sensors Journal, 24(6), 8701–8714.
[44] Reddy, V., Park, S., & Martínez, C. (2024). Few-shot adapters for cross-interrogator transfer. Sensors and Embedded AI, 12(3), 199–214.
[45] Reddy, V., Torres, B., & Zhao, D. (2024). Low-latency CNN–TCN hybrid models for subsea FBG demodulation. Robotics and Sensing Systems, 18(1), 66–85.
[46] Rossi, M., Singla, A., & Hernández, J. (2025). Edge-ready attention and calibration-aware FBG deployment. IEEE Embedded Systems Letters, 17(1), 15–22.
[47] Singh, R., Kumar, S., & Patel, A. (2023). Temporal convolutional networks for streaming FBG demodulation. IEEE Sensors Journal, 23(12), 13444–13456.
[48] Singla, A., Rossi, M., & Hernández, J. (2025). OT-aware few-shot learning for FBG interrogator adaptation. Optics Express, 33(4), 6123–6138.
[49] Liao, J., Chen, H., & Park, S. (2020). Uncertainty calibration (ECE) and robust regression for fiber sensing models. IEEE Transactions on Instrumentation and Measurement, 69(11), 8873–8888.
[50] Stevenson, R., Novak, T., & Borges, L. (2021). Memory-efficient uncertainty tracking in embedded FBG systems. Embedded AI for Photonics, 7(2), 89–106.
[51] Velasco, F., Martínez, C., & Chen, L. (2023). OSNR-aware synthetic augmentation for robust FBG decoding. Marine Photonic Analytics, 27(3), 140–160.
[52] Velasco, F., Park, S., & Rahman, M. (2023). Quantisation-aware training for embedded FBG demodulation engines. Sensors and Signal Processing Systems, 21(2), 231–248.
[53] Zhu, H., Álvarez, P., & Chen, L. (2021). Reject–verify uncertainty gates for subsea fiber sensing. IEEE Sensors Letters, 5(4), 1–4.
[54] Zhu, H., Ortiz, T., & Rahman, M. (2021). Evidential regression for calibrated multi-parameter FBG demodulation. Optical Machine Learning, 8(3), 50–68.
[55] Ibrahim, A., Hsu, Y., & Borges, L. (2024). Domain adaptation and few-shot recalibration for FBG interrogators. *IEEE Internet of Things Journal, 11(8), 14222–14235.
[56] Okonkwo, E., Petrov, A., & Singh, R. (2022). Motion-phase conditioning for streaming FBG demodulation. *Real-Time Optical Systems, 9(3), 67–84.
[57] Park, S., Chen, L., & Rahman, M. (2025). Physics-guided ML guardrails for subsea fiber sensing. IEEE Access, 13, 15522–15540.
[58] Park, S., Rahman, M., & Álvarez, P. (2025). Motion-phase conditioning for streaming demodulation in ROVs. Marine Robotics Journal, 9(1), 82–104.
[59] Zhao, D., Ortiz, T., & Li, H. (2024). Physics-guided simulation for subsea FBG interpretation. Sensors and Structured Environments, 32(1), 143–165.
[60] Zhao, D., Park, S., & Reddy, V. (2024). Spectral-derivative features and compact indices for low-OSNR decoding. *Journal of Optical Engineering, 61(2), 201–220.
[61] Okeke, R. O., Idigo, V. E., Akemi, M. O., & Ogbuokebe, S. K. (2021). Passive optical network, a fibre to the “X” approach. European Journal of Engineering and Technology Research, 6(3), 66–74.
[62] Okeke, R. O., Nnamdi, U. C., & Onu, P. I. (2022). Dispersion compensation algorithm for single mode fiber. European Journal of Engineering and Technology Research, 7(2), 92–98.
How to cite this paper
@article{1717403,
author = {Agomuo Hyginus C., Okeke Remigius Obinna},
title = {A Systematic Review of Deep Learning Approaches for Fiber Bragg Grating Sensor Data Interpretation in Subsea Multi-Parameter Monitoring},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {9},
number = {11},
pages = {868-885},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1717403.pdf},
abstract = {Fiber Bragg Grating (FBG) sensors have recently been identified as one of the most important technologies to provide subsea monitoring owing to high sensitivity, Electromagnetic Interference (EMI) immunity and the ability to measure a number of parameters at once. Although FBG-based sensing solutions are very promising, there are numerous challenges associated with spectrum interpretation in subsea conditions. Particularly, temperature, salinity and pressure parameters are cross-sensitive, with spectral distortions (polarisation effects, birefringence, wavelength effects and noise) causing complex nonlinear dependencies that are intractable to traditional signal processing methods. Recent development in the field of deep learning demonstrates remarkable achievements in the modelling of complicated spectral patterns to do multi-parameter estimation. Various kinds of neural architecture, such as convolutional, recurrent, and attention-based neural networks are employed to learn the local and global interactions among the spectral signals. In addition, new learning methods like self-supervised learning, federated learning, and physics-aware modelling are designed to alleviate the issues such as data scarcity, domain variability, and deployment-related challenges. Nonetheless, the current deep learning approaches have several critical weaknesses, such as the lack of modelling cross-sensitivity effects, the lack of good generalisation when operating under realistic settings, poor resistance to uncertainties, and real-time and edge-deployment issues. This paper presents a review of our work on the use of deep learning to interpret FBG spectral data in detail. In particular, we derive a theoretical framework to explain the problem, create a taxonomy to analyse the state-of-the-art approaches, and find out significant gaps in research and future directions.},
keywords = {Fiber Bragg Grating (FBG), Deep Learning, Subsea Monitoring, Remotely Operated Vehicle (ROV), Multi-Parameter Sensing, Uncertainty Quantification, Edge Deployment, Federated Learning, Physics-Informed Neural Network.},
month = {May},
doi = {https://doi.org/10.64388/IREV9I11-1717403}
}