软件定义互连(SDI)关键技术研究

批准号:
61972380
项目类别:
面上项目
资助金额:
60.0 万元
负责人:
元国军
依托单位:
学科分类:
计算机网络
结题年份:
2023
批准年份:
2019
项目状态:
已结题
项目参与者:
元国军
国基评审专家1V1指导 中标率高出同行96.8%
结合最新热点,提供专业选题建议
深度指导申报书撰写,确保创新可行
指导项目中标800+,快速提高中标率
微信扫码咨询
中文摘要
随着大数据和人工智能技术的兴起,数据中心和高性能中心的应用负载特征呈现愈发明显的多样化趋势,这给互连网络的灵活性提出了更高的要求。现有的软件定义网络(SDN)虽然支持OSI模型中网络层之上的灵活可控,但无法实现物理层的资源灵活分配。软件定义互连(SDI)理念应运而生,旨在打破网络底层硬件的刚性结构,实现物理拓扑可重构、带宽可调整和协议可定义等目标。.相比于传统的电交换,光交换技术具备更强的灵活性,例如通过控制空间光交换系统中的光信号传输方向或者波长交换系统中的信号波长来实现逻辑光路的动态路由,这就为实现物理拓扑的软件定义提供了可能。本课题将重点研究支持动态重构的光交换网络,在不改变物理布局布线情况下,通过软件定义的方法来动态切换光路,实现SDI中的物理拓扑重构。本课题的研究主要包括可重构光网络物理结构、软件定义的控制方法和负载特征匹配三个方面。
英文摘要
With the rapid development of big data and artificial intelligence, the traffic patterns of applications in data centers and supercomputers present more and more diversified. Therefore flexible interconnections became the future development direction and it would be desirable to dynamically match the network to the specific application. Although existing software-defined network (SDN) has improved the flexibility above network layer in OSI model, it can’t support the flexible resource reallocation in physical layer. Software Defined Interconnect (SDI) is proposed to break the rigid structure of the network hardware, and support topology reconfiguration, bandwidth adjustment, interconnect protocol definition and so on..Optical switching presents more flexible and can provide more support for SDI compared to traditional electrical switching. For example logical paths can be reconfigured by optical space switching or wavelength routing. This project will focus on the research of reconfigurable optical network, in which the physical topology can be dynamically reconfigured from one to another by software-defined control plane. The research includes physical architecture, reconfiguration control and topology matching.
期刊论文列表
专著列表
科研奖励列表
会议论文列表
专利列表
A New Optoelectronic Hybrid Network Based on 2 Scheduling Optimization of Optical Links
一种基于光链路二次调度优化的新型光电混合网络
DOI:10.1109/tc.2021.3054308
发表时间:2021
期刊:IEEE Transactions on Computers
影响因子:3.7
作者:En Shao;Guangming Tan;Zhan Wang;Guojun Yuan;Zheng Cao;Ninghui Sun
通讯作者:Ninghui Sun
Fast and scalable all-optical network architecture for distributed deep learning
用于分布式深度学习的快速且可扩展的全光网络架构
DOI:--
发表时间:2024
期刊:Journal of Optical Communications and Networking
影响因子:5
作者:Wenzhe Li;Guojun Yuan;Zhan Wang;Guangming Tan;Peiheng Zhang;G. Rouskas
通讯作者:G. Rouskas
国内基金
海外基金
