Performance impact of the interactions between time -triggered and rate-constrained transmissions in TTEthernet

Switched Ethernet is becoming a de-facto standard in industrial and embedded networks. Many of today’s applications benefit from Ethernet’s high bandwidth, large frame size, multicast and routing capabilities through IP, and the availability of the standard TCP/IP protocols. There are however many variants of Switched Ethernet networks, just considering the MAC level mechanisms on the stations and communication switches. An important technology in that landscape is TTEthernet, standardized as SAE6802, which allows the transmission of both purely time-triggered (TT) traffic and sporadic (or rate-constrained - RC) traffic. To the best of our knowledge, the interactions between both classes of traffic have not been studied so far in realistic configurations. This work aims to shed some light on the kind of performances, in terms of latencies, jitters and useful bandwidth that can be expected from a mixed TT and RC configuration. The following issues will be answered in a quantified manner by sensitivity analysis: How do both classes of traffic interfere with each other? What are the typical worst-case latencies and useful bandwidth that can be expected for a RC stream for various TT traffic loads? What is the overall impact of TTEthernet integration policy for the RC traffic? This study builds on a worst-case traversal time analysis developed by the authors for SAE6802, and explores these questions by experiments performed configurations of various sizes.