WANG Z H,LIU Z X,XU D N,et al. A frame preemption mechanism based on two-level queue scheduling in time-sensitive network[J]. Microelectronics & Computer,2024,41(6):65-72. doi: 10.19304/J.ISSN1000-7180.2023.0350
Citation: WANG Z H,LIU Z X,XU D N,et al. A frame preemption mechanism based on two-level queue scheduling in time-sensitive network[J]. Microelectronics & Computer,2024,41(6):65-72. doi: 10.19304/J.ISSN1000-7180.2023.0350

A frame preemption mechanism based on two-level queue scheduling in time-sensitive network

  • As one of the core technologies of network application scenarios such as industrial Ethernet, unmanned driving and avionics, Ethernet-based Time Sensitive Network (TSN) can provide deterministic transmission services with low delay and low jitter, which has been widely concerned by the industry. TSN traffic scheduling is a key technology to solve the deterministic and bounded delay problem of frame transmission. The frame preemption technology ensures the ultra-low delay of high-priority real-time data in TSN by dividing the data with different real-time requirements and allowing the transmission of high-priority real-time data to interrupt the transmission of low-priority non-real-time data. However, the frame preemption mechanism sets a "guard band" to isolate the transmission of high and low priority data, resulting in idle network links in the guard band. Aiming at the problem of link bandwidth loss caused by guard band, this paper improves the frame preemption mechanism and proposes a frame preemption mechanism based on Two-Level Queue Scheduling (TLQS). By setting up two groups of queues, the non-real-time data scheduled by the gated Time-Aware Shaper (TAS) mechanism is diverted, stored and scheduled for arbitration, and the idle bandwidth of the guard band is used for transmission selection. Simulation result shows that compared with the standard frame preemption mechanism, the frame preemption mechanism based on TLQS can effectively improve the network bandwidth utilization on the basis of reducing the additional overhead caused by preemption.
  • loading

Catalog

    Turn off MathJax
    Article Contents

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return