| 摘要: |
| 卫星地面中继站能够有效辅助普通便携式移动终端进行高质量的卫星通信。在电力网络缺失的地区，地面中继站依靠自身配备的光伏和风能转化装置供电，具有不稳定性，其能效问题值得研究。针对中继站在能量不充分不稳定条件下的能效优化问题，提出一种基于双深度Q网络的卫星地面中继站节能休眠策略。首先根据地面中继站的能耗特点设计智能体的状态空间、动作空间和奖励函数，然后利用算法对其能耗和服务状态的价值学习、特征提取能力训练智能体的最佳休眠策略。仿真结果表明，相较于传统基于规则的休眠方法，所提出的策略在低负载场景下有效地节省了21.55%的非通信能耗，同时提高了中继站3.7%的整体能效，在保障用户服务质量的同时显著降低了能源消耗。 |
| 关键词: 卫星网络 地面中继站；智能休眠；能效优化；深度强化学习 |
| DOI：10.20079/j.issn.1001-893x.240523003 |
|
| 基金项目:科技部重大专项(2022YFE03050004)；人工智能四川省重点实验室开放基金项目(2021RZJ01)；四川轻化工大学研究生创新基金(Y2023300) |
|
| An Intelligent Sleep Strategy for Satellite Network Ground Relay Station |
| LI Longhui,LI Yitao |
| (a.School of Automation and Information Engineering;b.Artificial Intelligence Key Laboratory of Sichuan Province,Sichuan University of Science and Engineering,Yibin 644000,China) |
| Abstract: |
| Satellite ground relay stations can effectively support regular portable mobile terminals in high-quality satellite communications.In regions without power grids,ground relay stations rely on their own photovoltaic and wind energy conversion devices for power supply,which is unstable,and their energy efficiency issues are worth studying.The authors propose an energy-saving sleep strategy for satellite ground relay stations based on double deep Q networks to address the issue of energy efficiency optimization of relay stations under conditions of insufficient and unstable energy.First,the state space,action space,and reward function of the agent are designed according to the energy consumption characteristics of the ground relay station.Then,the value learning and feature extraction capabilities of the algorithm are used to train the agent’s optimal sleep strategy.Simulation results show that compared with the traditional rule-based sleep method,the proposed strategy effectively saves 21.55% of non-communication energy consumption in low-load scenarios and improves the overall energy efficiency of the relay station by 3.7%.It significantly reduces energy consumption while ensuring user service quality. |
| Key words: satellite network ground relay station intelligent sleep energy efficiency optimization deep reinforcement learning |
《电讯技术》编辑部 