%0 Journal Article
%T 卫星网络智能路由优化机制
Satellite Network Intelligent Routing Optimization Mechanism
%A 张宇
%J Software Engineering and Applications
%P 332-345
%@ 2325-2278
%D 2025
%I Hans Publishing
%R 10.12677/sea.2025.142030
%X 本文针对低轨道卫星网络中动态拓扑结构与多优先级业务需求不匹配的问题，提出了一种结合分级服务标识与深度强化学习的动态路由优化方案。该方案包含三个核心模块：首先，基于IPv6原生字段的零开销服务标识机制来区分不同优先级业务；接着，提出分层式流量感知模型(H-CMS)，用于低开销、高精度地测量多类业务流量；最后，双延迟深度确定性策略梯度(TD3)算法，用于实时生成最优路由决策。实验结果表明，与传统方法相比，本方案显著提高了网络吞吐量和带宽利用率，特别是在卫星切换和高负载情况下表现更为稳定，吞吐量整体提升8%，带宽利用率维持在75%左右，为低轨道卫星网络中的差异化服务提供了有效解决方案。
Aiming at the problem of mismatch between dynamic topology and multi-priority service requirements in low-orbit satellite networks, this paper proposes a dynamic routing optimization scheme combining hierarchical service identification and deep reinforcement learning. The scheme consists of three core modules: first, a zero-overhead service identification mechanism based on IPv6 native fields is used to distinguish services of different priorities; second, a hierarchical traffic-aware model (H-CMS) is proposed to measure multi-class service traffic with low overhead and high precision; finally, a double-delay deep deterministic policy gradient (TD3) algorithm is used to generate optimal routing decisions in real time. Experimental results show that compared with traditional methods, this scheme significantly improves network throughput and bandwidth utilization, especially in satellite switching and high-load conditions. The overall throughput is increased by 8%, and the bandwidth utilization is maintained at about 75%, providing an effective solution for differentiated services in low-orbit satellite networks.
%K 低轨道卫星网络，
%K 多业务需求，
%K 分级服务，
%K 流量感知，
%K 强化学习
Low-Orbit Satellite Network
%K Multi-Service Requirements
%K Hierarchical Services
%K Traffic-Aware
%K Reinforcement Learning
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=112656