شبکه کانولوشنی و یادگیری چندنمونه‌ای برای برآورد غیرتماسی ضربان قلب از اطلاعات حالت کانال در شرایط داده ‌محدود

نوع مقاله : مقاله پژوهشی

نویسندگان

1 دانشجوی دکتری مهندسی برق، دانشکده مهندسی برق، دانشگاه یزد، یزد، ایران

2 دانشیار، دانشکده مهندسی برق، گروه مخابرات، دانشگاه یزد، یزد، ایران

10.22034/abmir.2026.24171.1214

چکیده

پایش غیرتماسی ضربان قلب با استفاده از اطلاعات حالت کانال (CSI) سیگنال وای‌فای، به‌عنوان رویکردی نوظهور در حسگری زیست‌پزشکی، امکان نظارت پیوسته بر وضعیت فیزیولوژیکی افراد را بدون نیاز به تماس فیزیکی یا تجهیزات پوشیدنی فراهم می‌سازد. بااین‌حال، تغییرپذیری قابل‌توجه الگوهای CSI میان افراد مختلف و کمبود داده‌های برچسب‌دار، توسعه مدل‌های یادگیری عمیق تعمیم‌پذیر را با چالش جدی مواجه می‌کند. در این پژوهش، یک چارچوب مبتنی بر یادگیری داده‌محدود برای تخمین ضربان قلب از سیگنال‌های CSI ارائه می‌شود که از ساختارهای یادگیری شباهت‌محور (Siamese) و رابطه‌محور (Relation-based) بهره می‌گیرد. در این چارچوب، شبکه‌های عصبی کانولوشنی SE-DenseNet، SEResNet10 و MobileNetV2 به‌عنوان رمزگذار ویژگی به‌کار گرفته‌شده‌اند تا نمایش‌های تمایز بخش از سیگنال‌های زمانی و طیف‌نگارهای فرکانسی CSI استخراج شود. سپس، با استفاده از تعداد محدودی نمونه پشتیبان، مدل قادر است به‌صورت سریع با توزیع داده سوژه جدید تطبیق یابد، بدون آنکه دچار بیش‌برازش شود. با ارزیابی به روش LOSO کمترین میزان خطا bpm 38/1 در روش شباهت محور با رمزگذار ویژگی SE-DenseNet حاصل گردید. این طراحی امکان تحلیل نقش هم‌ترازی میان معماری شبکه، نمایش زمانی یا فرکانسی CSI و راهبرد یادگیری داده‌محدود را در توسعه سامانه‌های پایش غیرتماسی ضربان قلب فراهم می‌سازد.

کلیدواژه‌ها

موضوعات

عنوان مقاله [English]

Convolutional Neural Network and Few-Shot Learning for Non-Contact Heat Rate Estimation from Channel State Information in Data-Limited Scenarios

نویسندگان [English]

  • Alireza Abolghasemi 1
  • S.M.T Almodarresi 2
1 Ph.D. Candidate in Electrical Engineering, Department of Electrical Engineering, Yazd University, Yazd, Iran
2 Associate Professor, Department of Electrical Engineering, Yazd University, Yazd, Iran
چکیده [English]

Non-contact heart rate monitoring using Wi-Fi Channel State Information (CSI) has emerged as a promising approach in biomedical sensing, enabling continuous physiological monitoring without physical contact or wearable devices. However, significant inter-subject variability in CSI patterns and the scarcity of labeled data pose major challenges to developing accurate and generalizable deep learning models. In this paper, a few-shot learning–based framework is proposed for heart rate estimation from CSI signals, leveraging both similarity-based (Siamese) and relation-based learning paradigms. Within this framework, convolutional neural networks including SE-DenseNet, SE-ResNet10, and MobileNetV2 are employed as feature encoders to extract discriminative representations from temporal CSI signals and frequency-domain CSI spectrograms. Using only a limited number of support samples, the proposed model is able to rapidly adapt to the data distribution of unseen subjects while effectively mitigating overfitting. Performance evaluation under the Leave-One-Subject-Out (LOSO) protocol demonstrates that the similarity-based model with an SE-DenseNet feature encoder achieves the lowest mean absolute error of 1.38 bpm. This study highlights the critical role of aligning network architecture, CSI representation (temporal or spectral), and few-shot learning strategy in the design of robust and non-contact heart rate monitoring systems.

کلیدواژه‌ها [English]

  • Heart Rate
  • Few-Shot Learning
  • Non-Contact Monitoring
  • Channel State Information (CSI)
  • Wi-Fi Sensing

مراجع

[1]     P. Susarla, A. H. Mirza, M. A. Rahman, and A. Bamis, “Non-contact multimodal indoor human monitoring systems: A survey,” Inf. Fusion., vol. 110, p. 102457, Oct. 2024, doi:10.1016/j.inffus.2024.102457.
[2]     D. Vandenberghe and J. Albrecht, “The financial burden of non-communicable diseases in the European Union: A systematic review,” Eur. J. Public Health., vol. 30, no. 4, pp. 833–839, Aug. 2020, doi:10.1093/eurpub/ckz073
[3]     M. I. Khan, M. R. Amin, M. S. Islam, and S. A. Rahman, “Tracking vital signs of a patient using channel state information and machine learning for a smart healthcare system,” Neural Comput. Appl., vol. 37, no. 28, pp. 23065–23079, Oct. 2025, doi:10.1007/s00521-020-05631-x.
[4]     M. Atif, S. Muralidharan, H. Ko, and B. Yoo, “COVID-Beat: A low-cost breath monitoring approach for people in quarantine during the pandemic,” J. Comput. Des. Eng., vol. 9, no. 3, pp. 992–1006, Jun. 2022, doi:10.1093/jcde/qwac037.
[5]     R. Yared and B. Abdulrazak, “Ambient technology to assist elderly people in indoor risks,” Computers, vol. 5, no. 4, p. 22, Oct. 2016, doi:10.3390/computers5040022.
[6]     S. Yue, H. He, H. Wang, H. Rahul, and D. Katabi, “Extracting multi-person respiration from entangled RF signals,” Proc. ACM Interact. Mob. Wearable Ubiquitous Technol., vol. 2, no. 2, Art. 86, Jul. 2018, doi:10.1145/3214289.
[7]     L. Qiao, Y. Zhang, Y. Liu, J. Wang, and Y. Li, “Spectral unmixing successive variational mode decomposition for robust vital signs detection using UWB radar,” IEEE Trans. Instrum. Meas., vol. 72, pp. 1–13, May. 2023, doi:10.1109/TIM.2023.3274171.
[8]     X. Wang, C. Yang, and S. Mao, “On CSI-based vital sign monitoring using commodity WiFi,” ACM Trans. Comput. Healthcare, vol. 1, no. 3, Art. 12, May. 2020, doi:10.1145/3377165.
[9]     J. C. H. Soto, I. Galdino, E. Caballero, V. Ferreira, D. Muchaluat-Saade, and C. Albuquerque, “A survey on vital signs monitoring based on Wi-Fi CSI data,” Comput. Commun., vol. 195, pp. 99–110, Nov. 2022, doi:10.1016/j.comcom.2022.08.004.
[10] J. Liu, Y. Wang, Y. Chen, J. Yang, X. Chen, and J. Cheng, “Tracking vital signs during sleep leveraging off-the-shelf WiFi,” in Proc. 16th ACM Int. Symp. Mobile Ad Hoc Netw. Comput., Hangzhou, China, Jun. 2015, pp. 267–276, doi:10.1145/2746285.2746303.
[11] R. Zhang, Y. Li, Z. Chen, H. Wang, and Q. Liu, “Hybrid subcarrier selection method for vital sign monitoring with long-term and short-term data considerations,” IEEE Sens. J., vol. 22, no. 23, pp. 23209–23220, Oct. 2022, doi:10.1109/JSEN.2022.3215443.
[12] B. Fang, N. D. Lane, M. Zhang, A. Boran, and F. Kawsar, “BodyScan: Enabling radio-based sensing on wearable devices for contactless activity and vital sign monitoring,” in Proc. 14th Annu. Int. Conf. Mobile Syst., Appl., Serv., Singapore, 2016, pp. 97–110, doi:10.1145/2906388.2906411.
[13] X. Wang, C. Yang, and S. Mao, “PhaseBeat: Exploiting CSI phase data for vital sign monitoring with commodity WiFi devices,” in Proc. IEEE 37th Int. Conf. Distrib. Comput. Syst. (ICDCS), 2017, pp. 1230–1239, doi:10.1109/ICDCS.2017.206.
[14] K. Tsubota, Y. Nagatsu, and H. Hashimoto, “Biometric information acquisition system using VMD in Wi-Fi channel status information,” in IECON 2021 – 47th Annu. Conf. IEEE Ind. Electron. Soc., 2021, pp. 1–6, doi:10.1109/IECON48115.2021.9589628.
[15] Y. Gu, X. Zhang, Z. Liu, and F. Ren, “WiFi-based real-time breathing and heart rate monitoring during sleep,” in Proc. IEEE Global Commun. Conf. (GLOBECOM), 2019, pp. 1–6, doi:10.1109/GLOBECOM38437.2019.9014297.
[16] [16] X. Zhang, Y. Wang, J. Liu, and D. Fang, “Wital: A COTS WiFi devices based vital signs monitoring system using NLOS sensing model,” IEEE Trans. Hum.-Mach. Syst., vol. 53, no. 3, pp. 629–641, Apr. 2023, doi:10.1109/THMS.2023.3264247.
[17] L. Gui, C. Ma, B. Sheng, Z. Guo, J. Cai, and F. Xiao, “In-home monitoring sleep turnover activities and breath rate via WiFi signals,” IEEE Syst. J., vol. 17, no. 2, pp. 2355–2365, Dec. 2023, doi:10.1109/JSYST.2022.3225072.
[18] J. A. Armenta-Garcia, F. F. Gonzalez-Navarro, J. Caro-Gutierrez, G. Galaviz-Yanez, J. E. Ibarra-Esquer, and W. Flores-Fuentes, “Mining Wi-Fi channel state information for breathing and heart rate classification,” Pervasive Mob. Comput., vol. 91, p. 101764, Apr. 2023, doi:10.1016/j.pmcj.2023.101768.
[19] J. Hu, J. Yang, J. B. Ong, D. Wang, and L. Xie, “ResFi: WiFi-enabled device-free respiration detection based on deep learning,” in Proc. IEEE 17th Int. Conf. Control & Autom. (ICCA), 2022, pp. 510–515, doi:10.1109/ICCA54724.2022.9831898.
[20] H. A. Solgun, A. Ozcan, and O. Pinarer, “Real-time heart rate monitoring via Wi-Fi signal,” in Proc. IEEE Int. Conf. Big Data, 2023, pp. 4973–4978, doi: 10.1109/BigData59044.2023.10386433.
[21] M. Liu, Z. Lin, P. Xiao, and W. Xiang, “Human biometric signals monitoring based on WiFi channel state information using deep learning,” arXiv preprint, arXiv:2203.03980, 2022. doi:10.48550/arXiv.2203.03980
[22] Z. Xie, H. Wang, S. Han, E. Schoenfeld, and F. Ye, “DeepVS: A deep learning approach for RF-based vital signs sensing,” in Proc. 13th ACM Int. Conf. Bioinformatics, Comput. Biol. Health Informatics, Northbrook, IL, USA, 2022, pp. 1–10, doi:10.1145/3535508.3545554.
[23] Islam, S. Akter, and T. F. Sanam, “Secure Tracking of Patient’s Vital Signs Using CSI-Based Homomorphic Encryption-Enabled Deep Learning Framework,” IEEE J. Biomed. Health Inform, vol. 30, no. 2, pp. 1314-1327, Aug. 2026, doi:10.1109/JBHI.2025.3601969.
[24] Z. Wang, J. Li, W. Wang, Z. Dong, Q. Zhang, and Y. Guo, “Review of few-shot learning application in CSI human sensing,” Artif Intell Rev, vol. 57, no. 8, p. 195, May 2024, doi:10.1007/s10462-024-10812-4.
[25] Y. Song, T. Wang, P. Cai, S. K. Mondal, and J. P. Sahoo, “A comprehensive survey of few-shot learning: Evolution, applications, challenges, and opportunities,” ACM Comput. Surv., vol. 55, no. 13s, Art. 271, Dec. 2023, doi:10.1145/3582688.
[26] F.-F. Li, R. Fergus, and P. Perona, “One-shot learning of object categories,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 28, no. 4, pp. 594-611, Apr. 2006, doi:10.1109/TPAMI.2006.79.
[27] F. Sung, Y. Yang, L. Zhang, T. Xiang, P. H. S. Torr, and T. M. Hospedales, “Learning to Compare: Relation Network for Few-Shot Learning,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit., Salt Lake City, UT, USA, Jun. 2018, pp. 1199–1208, doi:10.1109/CVPR.2018.00131.
[28] Z. Zhao, C. Tingwei, C. Zhijie, L. Xiaoyang, L. Hang, C. Qimei, and Z. Guangxu, “CrossFi: A Cross Domain Wi-Fi Sensing Framework Based on Siamese Network,” IEEE Internet Things J, vol. 12, no. 12, pp. 20138-20155, Jun. 2025, doi:10.1109/JIOT.2025.3542850.
[29] Z. Wang, O. Lei, S. Chen, H. Ding, G. Wang, and F. Wang, “SimID: Wi-Fi-Based Few-Shot Cross-Domain User Recognition with Identity Similarity Learning,” Sensors (Basel, Switzerland), vol. 25, Aug. 2025, doi:10.3390/s25165151
[30] J. Mei, L. Kong, and F. Jia, “SE-DenseNet: A Dynamic Dense Network with Squeeze-and-Excitation Module for Pneumonia Classification in Chest X-ray Images,” in Proc. 9th Int. Conf. Multimedia Image Process., Osaka, Japan, 2024.doi:10.1145/3665026.3665047.
پژوهش های نظری و کاربردی هوش ماشینی
دوره 4، شماره 1
بهار و تابستان 1405
اردیبهشت 1405
صفحه 77-94

فایل ها

هم رسانی

ارجاع به این مقاله

آمار