一种基于Sarsa算法的通联队列管控方法  

A Communication Queue Management Method Based on Sarsa A lgorithm

作　　者：吴鹏 张轩[1] 陈健 Wu Peng;Zhang Xuan;Chen Jian(The 28th Research Institute of China Electronics Technology Group Corporation,Nanjing 210007,China)

机构地区：[1]中国电子科技集团公司第28研究所,南京210007

出　　处：《信息化研究》2025年第1期43-47,共5页INFORMATIZATION RESEARCH

摘　　要：在荒野救援应用场景下,卫星通信、北斗用户机、短波电台、超短波电台等野外通信装备提供的时断时续、低带宽、高时延的弱连接通信条件远远低于稳定的商用通信网络环境。目前典型的商用传统队列管理方法主要关注和解决链路拥塞问题,对于因时断时续的网络通信条件导致救援信息被随机丢弃问题一直没有合适的解决方法。本文提出一种基于Sarsa算法的通联队列管控方法,该方法由通联队列控制策略、通联能力估计建模等内容组成,以平均端到端可达率为强化学习目标,能够自主提供弱连接通信条件下救援信息及时传递的最优解决方案,为解决荒野救援等应用场景下弱连接通信信息的及时共享问题提供了一种方法。In the application scenario of wilderness rescue,the intermittent,low bandwidth,and high la-tency weak connection communication conditions provided by outdoor communication equipment such as satellite communication,Beidou user machines,shortwave radios,and ultra-shortwave radios are far lower than stable commercial communication network environments.Typical commercial traditional queue management methods mainly focus on and solve the problem of link congestion at present,and there has been no suitable solution for the problem of rescue information being randomly discarded due to intermittent network communication condi-tions.This paper presents a communication queue management method based on Sarsa algorithm,which con-sists of communication queue control strategy,communication capability estimation modeling,etc.The rein-forcement learning objective is the average end-to-end reachability rate,which can autonomously provide the op-timal solution for timely transmission of rescue information under weak connection communication conditions.It provides a method for solving the problem of timely sharing under weak connection communication condi-tions.

关 键 词：弱连接通信 通联队列管控 通联能力估计模型 强化学习 

分 类 号：TN939.12[电子电信—信号与信息处理]