A Quantitative Analysis of Interfaces to Time-Triggered Communication Buses
Nodes connected to a time-triggered (TT) network can access the network interface in two different ways, synchronously or asynchronously, which greatly impacts communication timing and message lifespans (i.e., the time from writing a message to its send buffer till the time when the message is read by the receiver). In this paper we present a clear timing model to reason about the timing variation possible with TT interfaces. This model facilitates the quantitative analysis of the message lifespans of synchronous and asynchronous TT interfaces. Further, we develop a tool to search for node and network configurations that minimise or maximise message lifespans. We show that choosing the right configuration for synchronous interface access can reduce message lifespan significantly (we observed a factor of 9 even for small scenarios). While industrial practice typically is to choose a slot allocation a priory, we show that optimising the slot allocation in coordination with task scheduling gives an extra edge in obtaining minimal message lifespans. For nodes with synchronous interface access, the tool determines the parameters needed to obtain minimal message lifespan and jitter.