东北大学学报:自然科学版 ›› 2020, Vol. 41 ›› Issue (6): 784-791.DOI: 10.12068/j.issn.1005-3026.2020.06.005

• 信息与控制 • 上一篇    下一篇

SSA:一种面向CQF模型的TSN资源调度算法

姜旭艳, 严锦立, 全巍, 孙志刚   

  1. (国防科技大学 计算机学院, 湖南 长沙410073)
  • 收稿日期:2019-09-29 修回日期:2019-09-29 出版日期:2020-06-15 发布日期:2020-06-12
  • 通讯作者: 姜旭艳
  • 作者简介:姜旭艳(1998-),女,湖南永州人,国防科技大学博士研究生; 孙志刚(1973-),男,江苏连云港人,国防科技大学研究员.
  • 基金资助:
    国家重点研发计划项目(2018YFB1800402); 国家自然科学基金资助项目(61802417,61702538); 国防科技大学校内科研项目(ZK17-03-53,ZK18-03-40).

SSA:CQF-oriented Scheduling Algorithm in Time-Sensitive Networking

JIANG Xu-yan, YAN Jin-li, QUAN Wei, SUN Zhi-gang   

  1. College of Computer, National University of Defense Technology, Changsha 410073, China.
  • Received:2019-09-29 Revised:2019-09-29 Online:2020-06-15 Published:2020-06-12
  • Contact: JIANG Xu-yan
  • About author:-
  • Supported by:
    -

摘要: 时间敏感网络(time sensitive networking,TSN)的目标是在工业控制和5G领域为时间敏感流提供低延迟、低抖动的确定性传输服务.资源调度是保证TSN服务质量的关键技术之一.为此,TSN标准中提出循环队列转发模型(cyclic queuing and forwarding,CQF),但是缺乏相应的资源调度算法对多流进行合理的资源分配以避免传输冲突.由于时间敏感流量具有周期性、特征预知的特点,主要从时间维度研究如何将CQF模型中的队列资源合理分配给多条数据流.将该问题抽象为多约束条件下的资源规划最大化问题,提出基于起始时隙分配的轻量级资源调度算法(start-slot assignment based scheduling algorithm,SSA).SSA通过对端系统上发送时隙进行调节,在满足约束条件的前提下最大化对队列资源的使用,避免了端到端逐跳的时隙分配.实验结果表明,与不控制时隙的直接调度方式相比,该算法能够将成功调度的流数目平均提高41.84%.

关键词: 时间敏感网络, 确定性传输, 循环队列转发模型, 网络资源调度, 最大化问题

Abstract: Time-sensitive networking (TSN) is designed for providing determined transmission service with low latency and jitter in industrial control network and 5G.Resource scheduling, the core of TSN, is very crucial to guarantee the quality of service (QoS). Recently, TSN puts forward cyclic queuing and forwarding model (CQF). However, the corresponding scheduling algorithm is demanded for allocating resource rationally to avoid transmission collision.Since the TSN flow is periodic and predetermined, how to allocate resource for multiple TSN flows rationally from time dimension was researched. Resource allocation into multi-constrained maximization problem was abstracted. Then, a start-slot assignment based scheduling algorithm (SSA) was proposed. By adjusting the send slot at end system, SSA maximizes the queue resource utilization with meeting related constraints, which avoids complex time slot allocation hop by hop. Compared with the case without adjusting sending slot, experimental results demonstrate that SSA improves the number of flows scheduled successfully by 41.84% on average.

Key words: time-sensitive networking(TSN), deterministic transmission, cyclic queuing and forwarding model(CQF), network resource scheduling, maximization problem

中图分类号: