北京信息科技大学学报（自然科学版）2026，Vol.41Issue(4)：13-23,11.DOI:10.16508/j.cnki.11-5866/n.2026.04.002
基于GRU-MAPPO的应急通信无人机轨迹优化与功率分配
GRU-MAPPO-based trajectory optimization and power allocation for emergency communication UAVs
摘要
Abstract
In post-disaster scenarios with damaged terrestrial communication infrastructure,unmanned aerial vehicles(UAVs)can be rapidly deployed as aerial base stations to establish a temporary emergency communication network.However,the distribution of users in disaster areas is often significantly uneven,and the coverage capability of a single UAV is limited.This poses challenges for multi-agent systems to collaboratively ensure service quality in dynamic environments.To address this issue,a joint trajectory and resource optimization framework based on gated recurrent unit multi-agent proximal policy optimization(GRU-MAPPO)was proposed.In this framework,a GRU module was introduced into the process of policy learning,enabling UAVs to adaptively adapt their flight trajectories and transmit power under partial observable conditions,thereby improving system throughput and energy efficiency.To address the slow training convergence and low sample efficiency issues caused by the high-dimensional state space and continuous action space,a hybrid reward function was designed and combined with a linear curriculum learning scheduling strategy to improve the training process and facilitate convergence.Simulation results show that the proposed algorithm achieves better convergence stability and outperforms benchmark algorithms in terms of throughput and energy efficiency,providing enhanced reliability and environmental adaptability of communication systems in complex post-disaster rescue scenarios.关键词
无人机/应急通信/近端策略优化/课程学习/深度强化学习Key words
unmanned aerial vehicle(UAV)/emergency communication/proximal policy optimization(PPO)/curriculum learning/deep reinforcement learning分类
信息技术与安全科学引用本文复制引用
吕梁玮,谭玲玲,刘臻..基于GRU-MAPPO的应急通信无人机轨迹优化与功率分配[J].北京信息科技大学学报（自然科学版）,2026,41(4):13-23,11.基金项目
工信部高质量发展专项(DZBM2025-BJ-0025) （DZBM2025-BJ-0025）
