| 摘要: |
| 针对多波束低轨(Low Earth Orbit,LEO)卫星通信系统中动态干扰抑制、频谱效率提升及频繁切换带来的性能瓶颈问题,提出了一种基于深度强化学习的时空频多维资源协同优化方法。通过引入地面用户空间聚类算法实现动态波束覆盖重构,结合Acto-Critic架构的深度强化学习框架,利用近端策略优化算法实现波束激活、频谱分配与功率控制的协同优化。仿真结果表明,所提算法在400次迭代训练后收敛,相较于现有方案,系统平均吞吐量提升14.57%,星间切换次数降低43.79%,且干扰满足低于设定的干扰门限。 |
| 关键词: 空天地一体化网络 动态波束调度 资源分配 深度强化学习 |
| DOI:10.20079/j.issn.1001-893x.250802001 |
|
| 基金项目: |
|
| Beam Scheduling and Resource Co-optimization Method for LEO Satellites Based on DRL |
| LIU Motian,Yasenjiang Abudureheman,PAN Zhiwei,LUO Hao,MA Jianhui,QING Li,XUN Yuan |
| (1.Southwest China Institute of Electronic Technology,Chengdu 610036,China;2.School of Computer and Communication Engineering,University of Science and Technology Beijing,Beijing 100083,China) |
| Abstract: |
| To address the performance bottlenecks arising from dynamic interference suppression,spectral efficiency enhancement,and frequent handoffs in multi-beam low Earth orbit(LEO) satellite communication systems,a spatial-temporal frequency multidimensional resource collaborative optimization solution based on deep reinforcement learning is proposed.By introducing the ground user spatial clustering algorithm to achieve dynamic beam coverage reconstruction,combined with the deep reinforcement learning framework of Acto-Critic architecture,the proximal policy optimization algorithm is used to achieve collaborative optimization of beam activation,spectrum allocation,and power control.Simulation results show that the proposed algorithm converges after 400 training epochs.Compared with those of existing schemes,the average throughput of the system is improved by 14.57%,the number of inter satellite handoffs is reduced by 43.79%,and the interference is lower than the set threshold. |
| Key words: integrated space terrestrial network dynamic beam scheduling resource allocation deep reinforcement learning |