基于强化学习的资源最优化逻辑拓扑映射算法  

Algorithm of logical topology mapping for resource optimization based on reinforcement learning

作　　者：王亚男 杨雪[2] 庄浩涛 朱敏[2] 康乐 赵永利[3] WANG Ya'nan;YANG Xue;ZHUANG Haotao;ZHU Min;KANG Le;ZHAO Yongli(China Electric Power Research Institute,Beijing 100192,China;State Grid Sichuan Electric Power Company,Chengdu 610041,China;State Key Laboratory of Information Photonics and Optical Communications,Beijing University of Posts and Telecommunications,Beijing 100110,China)

机构地区：[1]中国电力科学研究院有限公司,北京100192 [2]国网四川省电力公司,成都610041 [3]北京邮电大学信息光子学与光通信国家重点实验室,北京100110

出　　处：《光通信技术》2020年第6期46-50,共5页Optical Communication Technology

基　　金：国家电网科技项目“电力通信光传输网络分布式仿真及优化技术研究”(5442XX180003)资助。

摘　　要：光传送网(OTN)中光节点波长复用/解复用器以及光开关矩阵可实现任意结构的逻辑拓扑在物理拓扑上的映射,不合理的映射方案将消耗额外端口资源。提出一种基于强化学习(RL)的逻辑拓扑最优化映射算法,将预处理后的拓扑状态和逻辑通道数据用于训练RL模型,以对逻辑通道进行全局波长资源分配,最终达到资源最优化目的。仿真结果表明:所提算法有效减小逻辑拓扑映射过程中的资源消耗,从而最小化网络部署成本。The optical node wavelength multiplexer/demultiplexer and optical switch matrix in optical transport network(OTN)can map the logical topology of any structure on the physical topology,the unreasonable mapping scheme will consume additional port resources.A logic topology optimization mapping algorithm based on reinforcement learning(RL)is proposed.The preprocessed topological state and logical channel data are used to train the RL model,so as to allocate the global wavelength resources to the logical channel,and finally achieve the purpose of resource optimization.Simulation results show that the proposed algorithm can effectively reduce the resource consumption in the process of logical topology mapping,thus minimizing the cost of network deployment.

关 键 词：光传送网 逻辑拓扑映射 强化学习 网络资源分配 

分 类 号：TN915[电子电信—通信与信息系统]