数据与模型协同驱动的智能光网络架构与关键技术  

Network architecture and key technologies of intelligent optical networks driven by data and model

作　　者：谷志群 张佳玮[1] 纪越峰[1] 于浩 塔里克·塔勒布 GU Zhiqun;ZHANG Jiawei;JI Yuefeng;YU Hao;Tarik Taleb(State Key Laboratory of Information Photonics and Optical Communications,Beijing University of Posts and Telecommunications,Beijing 100876,China;Oulu University,Oulu 90570,Finland)

机构地区：[1]北京邮电大学信息光子学与光通信国家重点实验室,北京100876 [2]奥卢大学无线通信中心,芬兰奥卢90570

出　　处：《电信科学》2022年第7期18-30,共13页Telecommunications Science

基　　金：国家重点研发计划项目(No.2018YFB1800802)。

摘　　要：网络的规模升级和超大连接、超高带宽、超低时延应用的不断深化,对光传输网络资源利用和网络差异化服务提出了更高要求,使得传统模型驱动下的网络形态和配置方式面临挑战。基于数据与模型协同驱动思想,提出“3层3循环”架构及其“3可功能”特征的智能光网络技术方案,并对智能化实现技术展开研究,通过开发设计的智能传输网络平台对所提算法的性能进行测试,经验证,数据与模型协同驱动的智能光网络传输性能得到有效提升,为实现网络智能化提供了理论技术支撑。With the expanding scale of the networks and the emerging applications of ultra-massive connection,ultra-high bandwidth and ultra-low delay,higher requirements are proposed for the utilization of optical transmission network resources and network differentiated services,which makes the network and configuration mode driven by the traditional approaches face challenges.Based on the idea of data and model collaborative driving,the intelligent optical network technologies of three-layerthree-cycle architecture and three functions were proposed,and the implementation technology was studied.The performance of the proposed algorithms was evaluated based on the designed intelligent transmission network platform.The validation results show that the transmission performance of intelligent optical network driven by collaborative data and model has been effectively improved,which provides theoretical and technical support for the realization of network intelligence.

关 键 词：智能光网络 数据与模型协同驱动 物理层状态感知 网络层资源联动 业务层切片定制 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]