Joint Optimization of 3D Trajectory and Resource Allocation in Multi-UAV Systems via Graph...

With their high mobility and ease of deployment, unmanned aerial vehicle (UAV)-assisted communication systems have emerged as a prominent area of academic research and a cornerstone technology for Sixth-Generation (6G) mobile communication networks. This paper investigates a multi-UAV downlink wireless communication system in which users exhibit random movement on the ground. To maximize the sum-rate of all users over the observation period, we propose a joint optimization framework that integrates user association, UAV 3D trajectory design, and power allocation, while addressing channel estimation across different timescales. In the long timescale, we model the UAV-user connections as a graph and utilize a graph neural network to jointly optimize user association and UAV trajectories. In the short timescale, we deploy a deep unfolding network for efficient channel estimation and power allocation. Simulation results validate the effectiveness of the proposed approach, showcasing significant performance improvements.

Joint Optimization of 3D Trajectory and Resource Allocation in Multi-UAV Systems via Graph Neural Networks

Jingwei Peng, Yunlong Cai, Zhejiang University; Jiantao Yuan, Hangzhou City University; Kai Ying, Shanghai Jiao Tong University; Rui Yin, Zhejiang University City College