On the Performance of Stream-based, Class-based Time-aware Shaping and Frame Preemption in TSN

Abstract

Time-sensitive Networking (TSN) is an evolving group of IEEE standards for deterministic real-time communication making standard Ethernet technology applicable to safetycritical application domains such as manufacturing or automotive systems. TSN includes several mechanisms influencing the timely forwarding of traffic, in particular, a time-triggered scheduling mechanism called time-aware shaper (TAS) and frame preemption to reduce the blocking time of high-priority traffic by low-priority traffic. Although these mechanisms have been standardized and products implementing them begin to enter the market, it is still hard for practitioners to select and apply suitable mechanisms fitting the problem at hand. For instance, TAS schedules can be calculated for individual streams or classes of traffic, and frame preemption with strict priority scheduling (w/o TAS) might seem to be an option in networks with extremely high data rates. In this paper, we make a first step towards assisting practitioners in making an informed decision when choosing between stream-based TAS, class-based TAS, and frame preemption by comparing these mechanisms in selected scenarios using our TSN network simulation tool NeSTiNg. Moreover, to facilitate the application of class-based TAS, we derive a formula for calculating class-based TAS configuration.

How to Optimize Joint Routing and Scheduling Models for TSN Using Integer Linear Programming

“…Noteworthy are also the schedulability drops for Confs. (8,9,16,17,21,25,26). In most of the cases, these drops can be explained by the increased execution times causing the solver to run out of time.…”

“…The switch assigns a frame to one of the queues by evaluating the Priority Code Point (PCP) in the VLAN tag of the frame and a configurable mapping of PCP values to the hardware queues. In our model, we take only high-priority streams into account which are scheduled on an individual basis, so-called stream-based scheduling [8]. The remaining traffic classes can be scheduled using different mechanisms, and the impact on these remaining classes is out of scope for this paper.…”

“…Theoretically, the number of start variables could also be reduced to a single start variable at the source vertex of the stream since our model forbids non-constant delays on the path (cf. no-wait-scheduling) [8]. However, such an optimization is not compatible with the way we build our base model and because we cannot reshape our equations to the start variable without getting dependencies to other start variables (cf.…”

Critical Offset Optimizations for Overlapping-Based Time-Triggered Windows in Time-Sensitive Network

“…Other window-based scheduling algorithms were presented in [5]- [8], [34], [35]. In [34], [35], the authors implemented their scheduling algorithms with no synchronization requirement for end systems. However, the prerequisite is that all non-terminal nodes must be synchronized with each other.…”

A Formal Method for Evaluating the Performance of TSN Traffic Shapers using UPPAAL

“…Although the above two approaches can calculate the worst-case delay by arithmetic, they rely too much on parameters and cannot automatically calculate. [14] used network simulation tool NeSTiNg to investigate the TSN scheduling mechanism. However, it's very time consuming and can not cover all cases.…”