A Configuration Framework for Multi-level Preemption Schemes in Time Sensitive Networking

Abstract

To reduce the latency of time-sensitive flows in Ethernet networks, the IEEE TSN Task Group introduced the IEEE 802.1Qbu Standard, which specifies a 1-level preemption scheme for IEEE 802.1 networks. Recently, serious limitations of this scheme w.r.t. flows responsiveness were exposed and the so-called multi-level preemption approach was proposed to address these drawbacks. As is the case with most, if not all, real-time and/or timesensitive preemptive systems, an appropriate priority-to-flow assignment policy plays a central role in the resulting performance of both 1-level and multi-level preemption schemes to avoid the over-provisioning and/or the suboptimal use of hardware resources. Yet on another front, the multi-level preemption scheme raises new configuration challenges. Specifically, the right number of preemption level(s) to enable for swift transmission of flows; and the flow-to-preemption-class assignment synthesis remain open problems. To the best of our knowledge, there is no prior work in the literature addressing these important challenges. In this work, we address these three challenges. We demonstrate the applicability of our proposed solution by using both synthetic and real-life use-cases. Our experimental results show that multi-level preemption schemes improve the schedulability of flows by over 12% as compared to a 1-level preemption scheme, and at a higher abstraction level, the proposed configuration framework improves the schedulability of flows by up to 6% as compared to the dominant Deadline Monotonic Priority Ordering.