Authors : H. A. Ramasombohitra; H. N. Ramanantsihoarana

Volume/Issue : Volume 10 - 2025, Issue 12 - December

Google Scholar : https://tinyurl.com/2anhafps

Scribd : https://tinyurl.com/5n89t9d8

DOI : https://doi.org/10.38124/ijisrt/25dec1251

Note : A published paper may take 4-5 working days from the publication date to appear in PlumX Metrics, Semantic Scholar, and ResearchGate.

Abstract : Portable medical devices designed for continuous monitoring of physiological parameters face a critical trade-off between energy autonomy and monitoring quality. This study presents an adaptive energy management approach based on the Pareto Set Learning - Multi-Objective Reinforcement Learning (PSL-MORL) algorithm for a system powered by photovoltaic solar energy recovery. The proposed approach dynamically generates Duty Cycle (DC) modulated according to the wearer's National Early Warning Score 2 (NEWS2), simultaneously targeting 24-hour energy neutrality operational (ENO) and compliance with clinical recommendations. A comparative study using MATLAB simulation evaluates three strategies: the PSL-MORL algorithm, an intensive monitoring policy (D = 14.3%), and an energy-saving policy (DC = 5.26%). The results demonstrate that the proposed approach guarantees 24-hours autonomy (+42.9% vs. intensive policy), improves the average NEWS2 compliance reward by 58.4% compared to the energy-saving policy, and optimizes the allocation of 142 measurements over clinically critical periods. These performances validate the effectiveness of the PSL- MORL algorithm in reconciling the energy and clinical constraints of autonomous portable medical devices.

Keywords : Adaptive Power Management, Energy-Aware Scheduling, IoWT, MORL, Pareto Set Learning, Physiological Monitoring.

References :

1. A. Esteva et al., « A guide to deep learning in healthcare », Nat. Med., vol. 25, no 1, p. 24‑29, janv. 2019, doi: 10.1038/s41591-018-0316-z.
2. E. J. Topol, « High-performance medicine: the convergence of human and artificial intelligence », Nat. Med., vol. 25, no 1, p. 44‑56, janv. 2019, doi: 10.1038/s41591-018-0300-7.
3. N. Mehta et A. Pandit, « Concurrence of big data analytics and healthcare: A systematic review », Int. J. Med. Inf., vol. 114, p. 57‑65, juin 2018, doi: 10.1016/j.ijmedinf.2018.03.013.
4. P. Szántó, T. Kiss, et K. J. Sipos, « Energy-efficient AI at the Edge », in 2022 11th Mediterranean Conference on Embedded Computing (MECO), juin 2022, p. 1‑6. doi: 10.1109/MECO55406.2022.9797178.
5. A. Ometov et al., « A Survey on Wearable Technology: History, State-of-the-Art and Current Challenges », Comput. Netw., vol. 193, p. 108074, juill. 2021, doi: 10.1016/j.comnet.2021.108074.
6. W. B. Qaim et al., « Towards Energy Efficiency in the Internet of Wearable Things: A Systematic Review », IEEE Access, vol. 8, p. 175412‑175435, 2020, doi: 10.1109/ACCESS.2020.3025270.
7. F. K. Shaikh et S. Zeadally, « Energy harvesting in wireless sensor networks: A comprehensive review », Renew. Sustain. Energy Rev., vol. 55, p. 1041‑1054, mars 2016, doi: 10.1016/j.rser.2015.11.010.
8. S. Shresthamali, M. Kondo, et H. Nakamura, « Adaptive Power Management in Solar Energy Harvesting Sensor Node Using Reinforcement Learning », ACM Trans. Embed. Comput. Syst., vol. 16, no 5s, p. 181:1-181:21, sept. 2017, doi: 10.1145/3126495.
9. O. Veligorskyi, M. Khomenko, R. Chakirov, et Y. Vagapov, « Performance analysis of a wearable photovoltaic system », in 2018 IEEE International Conference on Industrial Electronics for Sustainable Energy Systems (IESES), janv. 2018, p. 376‑381. doi: 10.1109/IESES.2018.8349905.
10. J. Ortegón-Aguilar et al., « Multimodal Power Management Based on Decision Tree for Internet of Wearable Things Systems », Appl. Sci., vol. 13, no 7, Art. no 7, janv. 2023, doi: 10.3390/app13074351.
11. « nRF52840 ». Consulté le: 22 novembre 2025. [En ligne]. Disponible sur: https://www.nordicsemi.com/-/media/Software-and-other-downloads/Product-Briefs/nRF52840-Dongle-product-brief.pdf
12. S. Mohsen, A. Zekry, K. Youssef, et M. Abouelatta, « A Self-powered Wearable Wireless Sensor System Powered by a Hybrid Energy Harvester for Healthcare Applications », Wirel. Pers. Commun., vol. 116, no 4, p. 3143‑3164, févr. 2021, doi: 10.1007/s11277-020-07840-y.
13. O. B. Akan, O. Cetinkaya, C. Koca, et M. Ozger, « Internet of Hybrid Energy Harvesting Things », IEEE Internet Things J., vol. 5, no 2, p. 736‑746, avr. 2018, doi: 10.1109/JIOT.2017.2742663.
14. R. D. Middlebrook et S. Cuk, « A general unified approach to modelling switching-converter power stages », in 1976 IEEE Power Electronics Specialists Conference, juin 1976, p. 18‑34. doi: 10.1109/PESC.1976.7072895.
15. D. Hussein, G. Bhat, et J. R. Doppa, « Adaptive Energy Management for Self-Sustainable Wearables in Mobile Health », Proc. AAAI Conf. Artif. Intell., vol. 36, no 11, Art. no 11, juin 2022, doi: 10.1609/aaai.v36i11.21451.
16. B. Martinez, M. Monton, I. Vilajosana, et J. D. Prades, « The Power of Models: Modeling Power Consumption for IoT Devices », IEEE Sens. J., vol. 15, no 10, p. 5777‑5789, oct. 2015, doi: 10.1109/JSEN.2015.2445094.
17. C. F. Hayes et al., « A practical guide to multi-objective reinforcement learning and planning », Auton. Agents Multi-Agent Syst., vol. 36, no 1, p. 26, avr. 2022, doi: 10.1007/s10458-022-09552-y.
18. Y. Rioual, Y. Le Moullec, J. Laurent, M. Khan, et J.-P. Diguet, Design and Comparison of Reward Functions in Reinforcement Learning for Energy Management of Sensor Nodes. 2021. doi: 10.48550/arXiv.2106.01114.
19. E. Liu et al., « Pareto set learning for multi-objective reinforcement learning », in Proceedings of the Thirty-Ninth AAAI Conference on Artificial Intelligence and Thirty-Seventh Conference on Innovative Applications of Artificial Intelligence and Fifteenth Symposium on Educational Advances in Artificial Intelligence, in AAAI’25/IAAI’25/EAAI’25, vol. 39. AAAI Press, févr. 2025, p. 18789‑18797. doi: 10.1609/aaai.v39i18.34068.
20. C. Qian, Y. Yu, et Z.-H. Zhou, « On constrained boolean pareto optimization », in Proceedings of the 24th International Conference on Artificial Intelligence, in IJCAI’15. Buenos Aires, Argentina: AAAI Press, juill. 2015, p. 389‑395.
21. X. Lin, Z. Yang, et Q. Zhang, « Pareto Set Learning for Neural Multi-Objective Combinatorial Optimization », présenté à International Conference on Learning Representations, oct. 2021.
22. D. M. Roijers, P. Vamplew, S. Whiteson, et R. Dazeley, « A Survey of Multi-Objective Sequential Decision-Making », J. Artif. Intell. Res., vol. 48, p. 67‑113, oct. 2013, doi: 10.1613/jair.3987.
23. K. Van Moffaert, M. M. Drugan, et A. Nowé, « Scalarized multi-objective reinforcement learning: Novel design techniques », in 2013 IEEE Symposium on Adaptive Dynamic Programming and Reinforcement Learning (ADPRL), avr. 2013, p. 191‑199. doi: 10.1109/ADPRL.2013.6615007.
24. S. Natarajan et P. Tadepalli, « Dynamic preferences in multi-criteria reinforcement learning », in Proceedings of the 22nd international conference on Machine learning, in ICML ’05. New York, NY, USA: Association for Computing Machinery, août 2005, p. 601‑608. doi: 10.1145/1102351.1102427.
25. C. Liu, X. Xu, et D. Hu, « Multiobjective Reinforcement Learning: A Comprehensive Overview », IEEE Trans. Syst. Man Cybern. Syst., vol. 45, no 3, p. 385‑398, mars 2015, doi: 10.1109/TSMC.2014.2358639.
26. P. Vamplew, R. Dazeley, A. Berry, R. Issabekov, et E. Dekker, « Empirical evaluation methods for multiobjective reinforcement learning algorithms », Mach Learn, vol. 84, no 1‑2, p. 51‑80, juill. 2011, doi: 10.1007/s10994-010-5232-5.