Minimizing the End-to-End Latency in Multi-Segment Time-Triggered Networks

Distributed applications complexity is steadily increasing. A well-known technique used to manage such complexity consists in decomposing the whole system in different quasi-independent subsystems, which are also frequently distributed. Inter-subsystem communication, when necessary, is performed via gateway nodes that filter in and outgoing traffic. For real-time systems, this architecture poses additional design challenges, since it becomes necessary to consider both intra and inter-network message exchanges with real-time constraints. This work addresses multi-segment time-triggered networks, and, for this class of systems, presents a methodology, which allows minimizing the end-to-end delay for the inter-network traffic. This goal is achieved by bounding the transmission windows of the messages in each one of the intervening networks, and then, taking advantage of the properties of timetriggered systems, making such transmission windows starting as soon as possible. In particular, this is based on the determination of the phase of a message that allows maximizing its best-case response time.