.png){kind=icon}
.png){kind=icon}
Return
Secrecy energy efficiency optimization for multi-UAV air-to-ground communications under uncertain eavesdropping
.png){kind=icon}
PRE
AI
薛
L
Z
S
DOI：10.1016/j.phycom.2026.103093.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In 6G-oriented air-ground integrated networks, secure cooperative transmission among multiple unmanned aerial vehicles (UAVs) in urban critical infrastructure scenarios faces two fundamental challenges. First, the locations of eavesdroppers are difficult to estimate accurately, rendering secrecy evaluation and security guarantee unreliable. Second, the strong coupling between propulsion energy consumption and communicationrelated energy expenditure causes high secrecy performance to be achieved at the cost of substantial energy consumption. To address these challenges, this paper proposes a joint secrecy-energy utility maximization framework under eavesdropper location uncertainty and UAV energy constraints. Specifically, a cooperative system model consisting of transmitting UAVs and jamming UAVs is established, where three-dimensional (3D) trajectory planning, subchannel selection, and communication/jamming power allocation are jointly optimized over discrete time slots. A circular uncertainty region is employed to characterize the eavesdropper location error, based on which an upper bound on the worst-case signal-to-interference-plus-noise ratio (SINR) of the eavesdropping link is derived, leading to a verifiable lower bound (LB) on the robust secrecy rate. Furthermore, both UAV propulsion energy consumption and communication and jamming energy consumption are incorporated into a long-term utility function to capture the tradeoff between secrecy performance and energy efficiency. Under the centralized training and decentralized execution (CTDE) framework, a multi-agent proximal policy optimization (MAPPO) algorithm is developed to enable cooperative online decision-making among multiple UAVs. Simulation results demonstrate that, compared with representative reinforcement learning baseline algorithms, the proposed approach achieves faster convergence and superior secrecy energy efficiency (SEE), while exhibiting enhanced robustness against eavesdropper location uncertainty.
Keywords:
Multi-UAV cooperation
Physical layer security
Secrecy energy efficiency optimization
Three-dimensional trajectory planning
Deep reinforcement learning
Journal
IF:
2.2
Papers:
279
Citations:
2.6K
Organization
Cited Papers