Citation

Ali, Farman and Khan, Irfan Ullah and Roslee, Mardeni and Ullah, Yasir and Ismail, Azmi and O.A, Idris (2025) Deep Graph Reinforcement Learning for UAV-to-Ground Secure Communication with Height-Dependent Line-of-Sight Estimation in Urban Scenarios. In: 2025 International Conference on Engineering and Emerging Technologies (ICEET), 22-23 October 2025, Kuala Lumpur, Malaysia.

Text 31.pdf - Published Version Restricted to Repository staff only Download (1MB)

Abstract

This paper proposes a deep graph reinforcement learning (DGRL) framework for U2G secure communication, incorporating height-dependent LoS estimation into trajectory and security optimization. The proposed system models the UAV’s six-dimensional (6D) mobility while capturing spatial interactions between users, obstacles, and unauthorized receivers (UR) using a graph neural network (GNN). A soft actor–critic (SAC) RL algorithm is applied to optimize the UAV’s trajectory with maximizing the secrecy rate. The LoS probability is mathematically expressed as a function of UAV height and elevation angle, calibrated using ITU-R and 3GPP TR 38.901 specifications. To validate the framework, a ray tracing (RT) simulation environment is developed, modeling realistic urban scenario and multipath propagation based on ITU and 3GPP guidelines. Simulation results demonstrate that the proposed method enhances secure coverage, LoS reliability, and trajectory efficiency compared to baselines.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	UAV-to-ground communication, Deep graph reinforcement learning, Unauthorized reciever detection, 6D trajectory optimization, LoS probability modeling
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101-6720 Telecommunication. Including telegraphy, telephone, radio, radar, television
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	20 Apr 2026 04:47
Last Modified:	20 Apr 2026 04:47
URII:	http://shdl.mmu.edu.my/id/eprint/15792

Downloads

Edit (login required)