Toward a Scalable Trajectory Approach for FIFO Scheduled Flows Sharing a Switched Ethernet network

Real-time systems are highly complex and distributed. One of the major challenges in the design process of these systems is timing analysis. We are interested in an analytical method called the Trajectory Approach (TA). It allows upper bounding the end-to-end response time of any flow exchanged across a studied system. In particular, we are investigating the scalability of this approach when applied on large industrial networks. In this paper, we show on a simple example why the computation of the upper bound can be complex. We also prove that for flows satisfying a specific condition, the calculation of the end-to-end response time is extremely fast. This reduce the total runtime required to analyze the underlying network. Finally, the approach is applied on two different size networks, the simulation results show the scalability of our solution without significant loss of the determined upper bound.