Machine Learning-Based Power Allocation for Covert Communication in LEO Satellite–UAV Cooperative Networks
Minjeong Kang1, Jung Hoon Lee1,*, Il-Gu Lee2,*
1 Department of Electronics Engineering and Applied Communications Research Center, Hankuk University of Foreign Studies, Yongin, Republic of Korea
2 Department of Future Convergence Technology Engineering, Sungshin Women’s University, Seoul, Republic of Korea
* Corresponding Author: Jung Hoon Lee. Email: 
; Il-Gu Lee. Email: 
(This article belongs to the Special Issue: Advanced Security and Privacy for Future Mobile Internet and Convergence Applications: A Computer Modeling Approach)
Computer Modeling in Engineering & Sciences https://doi.org/10.32604/cmes.2026.078247
Received 27 December 2025; Accepted 02 March 2026; Published online 17 March 2026
Abstract
In next-generation non-terrestrial network environments, the increasing risk of detection by unauthorized observers has motivated extensive research on covert communication approaches that minimize the probability of detection. In particular, jamming-assisted cooperative covert communication has attracted significant attention as an effective approach to simultaneously ensure communication performance and security, leading to growing interest in cooperative architectures among heterogeneous platforms. This study investigates covert communication in Low Earth Orbit (LEO) satellite–unmanned aerial vehicle (UAV) cooperative networks, where the LEO satellite serves a legitimate user, while the UAV acts as a cooperative jammer to enhance covertness. A network that integrates a LEO satellite with wide service coverage and a UAV with high mobility offers flexible support for covert communication in diverse environments. However, the problem of optimally allocating power between the LEO satellite and UAV while satisfying the covert communication constraint inherently exhibits a non-convex structure, which commonly necessitates a discretized grid-search baseline over feasible candidate combinations. As the number of candidates increases, this approach suffers from rapidly increasing computational complexity. To address this computational burden, this study proposes a machine-learning (ML)–based power-allocation scheme. The proposed ML model leverages key channel-related and covertness-related features to efficiently select an effective pair of power scaling factors, while significantly reduced computational complexity. Simulation results demonstrate that the proposed scheme achieves comparable average covert rate to that of the discretized grid-search baseline while requiring substantially lower computational complexity. These results further indicate that the proposed scheme enables low-latency and efficient power control in LEO satellite–UAV cooperative networks. Finally, future work will extend the proposed scheme to more complex multi-LEO satellite–UAV cooperative scenarios through joint optimization of additional system parameters.
Keywords
Low earth orbit (LEO) satellite communications; unmanned aerial vehicle (UAV) cooperative jamming; covert communication; power allocation; machine learning (ML)
Open Access
Login
Register
Submit a Paper
Propose a Special lssue
Download PDF
Downloads
Citation Tools
