Distributed Game-Theoretic Learning for Integrated Sensing and Communication in Multi-UAV Systems: A Two-Stage MADRL Approach
IEEE Transactions on Mobile Computing · 2026-12 · 25 (12) · 24444-24461
DOI: 10.1109/tmc.2026.3735303 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Mobile Computing, 2026-12. Verfasst von Junxu Cheng und 5 weiteren Autor:innen. Die Arbeit nennt 26 Literaturangaben. Gefördert von: National Natural Science Foundation of China, National Research Foundation, Singapore and Infocomm Media Development Authority through its Communications and Connectivity Bridging Funding Initiative.
Abstract (Originalsprache)
The integrated sensing and communication (ISAC)-enabled unmanned aerial vehicless (UAVss) have garnered significant attention for their flexible observation and enhanced communication capabilities. However, existing works on ISAC-enabled multi-UAVs systems predominantly rely on centralized solutions, which assume full access to global information. In practice, such an assumption is impractical due to the limited communication and sensing ranges of UAVss, making centralized solutions less feasible. To overcome this limitation, we focus on a distributed scheme in which each UAVs in a multi-UAVs system can only utilize local information to make decisions. Therefore, we aim to maximize sensing and tracking performance of the system via jointly optimizing selection of UAVs tracking targets, UAVs trajectory, and beamforming vector, and furthermore the maximization problem can be decomposed into several sub-problems with respect to the sensing and tracking performance of each UAVs. Due to the NP-hardness arising from non-convexity and coupling, we solve the problem by formulating a sensing and tracking performance maximization game. Based on the designed reward and potential functions, we show that the instantaneous game under a fixed decision time slot admits a pure Nash equilibrium, and then develop a distributed learning algorithm, named two-stage game-theoretic Nash learning (TSGTNL), to learn adaptive policies that respond to the state dependent equilibrium structure using only local information. Extensive simulations illustrate that the TSGTNL significantly improves the sensing and tracking performance in ISAC-enabled multi-UAVs systems, compared to representative methods, while relying solely on local information.
Förderung
- National Natural Science Foundation of China
- National Research Foundation
- Singapore and Infocomm Media Development Authority through its Communications and Connectivity Bridging Funding Initiative
Zitieren
APA
Cheng, J., Wang, K., Sun, Y., Shin, H., Quek, T. Q. S., & Win, M. (2026). Distributed Game-Theoretic Learning for Integrated Sensing and Communication in Multi-UAV Systems: A Two-Stage MADRL Approach. IEEE Transactions on Mobile Computing, 25(12), 24444-24461. https://doi.org/10.1109/tmc.2026.3735303
Vancouver
Cheng J, Wang K, Sun Y, Shin H, Quek TQS, Win M. Distributed Game-Theoretic Learning for Integrated Sensing and Communication in Multi-UAV Systems: A Two-Stage MADRL Approach. IEEE Transactions on Mobile Computing. 2026;25(12):24444-24461. doi:10.1109/tmc.2026.3735303
BibTeX
@article{Cheng2026,
author = {Cheng, Junxu and Wang, Kehao and Sun, Yingzhao and Shin, Hyundong and Quek, Tony Q. S. and Win, Moe},
title = {Distributed Game-Theoretic Learning for Integrated Sensing and Communication in Multi-UAV Systems: A Two-Stage MADRL Approach},
journal = {IEEE Transactions on Mobile Computing},
year = {2026},
volume = {25},
number = {12},
pages = {24444--24461},
doi = {10.1109/tmc.2026.3735303}
}
EUVIMED