A1 Journal article (refereed)
Game-Theoretic Power Allocation and Client Selection for Privacy-Preserving Federated Learning in IoMT (2025)
Liu, J., Chang, Z., Ye, C., Mumtaz, S., & Hämäläinen, T. (2025). Game-Theoretic Power Allocation and Client Selection for Privacy-Preserving Federated Learning in IoMT. IEEE Transactions on Communications, Early Access. https://doi.org/10.1109/tcomm.2024.3523968
JYU authors or editors
Publication details
All authors or editors: Liu, Jingyuan; Chang, Zheng; Ye, Chaoxiong; Mumtaz, Shahid; Hämäläinen, Timo
Journal or series: IEEE Transactions on Communications
ISSN: 0090-6778
eISSN: 1558-0857
Publication year: 2025
Volume: Early Access
Publisher: Institute of Electrical and Electronics Engineers (IEEE)
Publication country: United States
Publication language: English
DOI: https://doi.org/10.1109/tcomm.2024.3523968
Publication open access: Not open
Publication channel open access:
Abstract
In recent years, the Internet of Medical Things (IoMT) has significantly boosted the healthcare industry. Federated learning (FL) can enhance the utilization of patient data while protecting privacy. Despite the great potential of FL to enhance the architecture of IoMT, the need for effective interference management and the limited energy resources of IoMT devices make the integration of FL into IoMT environments particularly challenging. This study proposes an innovative framework to address these challenges by optimizing power allocation and client selection across participating IoMT devices in the FL process. By employing a Stackelberg game model, our approach orchestrates power allocation among IoMT devices to enhance communication efficiency while adhering to strict differential privacy (DP) standards. Regarding the availability of network state information, we propose non-uniform pricing and uniform pricing strategies, respectively. Then, we derive the optimal interference price and power for the IoMT devices using nonlinear programming and convex optimization. To tackle the issue of energy constraints in IoMT devices, we adopt Lyapunov optimization for adaptive client selection, ensuring sustainable device participation in the FL process over time. In addition, our approach integrates DP to protect patient data, carefully balancing between privacy and the accuracy of the learning model. Our extensive simulations demonstrate marked improvements in privacy preservation, communication efficiency, and energy management efficiency, highlighting the effectiveness of our proposed method over existing solutions.
Keywords: data models; game theory; resource allocation
Free keywords: federated learning; internet of medical things (IoMT); differential privacy; game theory; client selection; resource allocation
Contributing organizations
Related projects
- Autonomous Vehicular Edge Computing and Networking for
Intelligent Transportation- Hämäläinen, Timo
- European Commission
- Human-guided collAboRative Multi-Objective design of explaiNable,
faIr and privaCy-preserving AI for digital health- Hämäläinen, Timo
- European Commission
Ministry reporting: Yes
VIRTA submission year: 2025
Preliminary JUFO rating: 3