面向人工智能的数据通信网络发展  

Data Communication Network Development for Artificial Intelligence

作　　者：高巍 高静 杨哲 GAO Wei;GAO Jing;YANG Zhe(China Academy of Information and Communications Technology,Beijing 100083,China)

机构地区：[1]中国信息通信研究院,中国北京100083

出　　处：《中兴通讯技术》2024年第6期3-9,共7页ZTE Technology Journal

摘　　要：基于人工智能技术与业务对数据通信网络的需求,分析现有网络面向数据入算、智算中心互联、大规模AI训练3类场景时存在的问题,阐述“入算”“算内”“算间”网络关键技术创新情况,包括入算网络的业务创新探索,算内网络围绕架构以太网技术等多方面的革新,以及算间网络从IT、IP、光层开展的技术改进,并提出包含运营层、网络管控层、业务连接层、物理网络层的4层网络架构以优化数据通信网络。认为合理推动产业发展需有序规划标准化研究工作,递进式开展关键技术试点验证。Based on the requirements of artificial intelligence(AI)technology and business on data communication networks,this paper first analyzes the problems of the existing networks in three scenarios:data access for computing,interconnection of intelligent computing cen-ters,and large-scale AI training.The key technology innovation of"access-artificial intelligence data center","inter-artificial intelligence data center"and"intra-artificial intelligence data center"networks are then illustrated,including the business innovation exploration of"access-artificial intelligence data center"networks,the architecture innovation of"inter-artificial intelligence data center"networks around the Ether-net technology,and the technical improvement of"intra-artificial intelligence data center"networks from IT,IP,and the optical layer.After that,a 4-layer network architecture is proposed,including the operation layer,network control layer,service connection layer,and physical network layer,to optimize the data communication network.We believe that reasonable promotion of industrial development requires or-derly planning of standardization research and progressive pilot verification of key technologies.

关 键 词：人工智能 数据通信网络 入算网络 算间网络 算内网络 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]