Time-Triggered vs. Event-Triggered: A matter of configuration?

During the development of real-time systems one has either to plump for a time-triggered or an event-triggered architecture. Actually this decision deals with a non-functional property of a real-time system and should therefore be postponed as far as possible. Unfortunately, this property also exhibits functional qualities during the development of real-time systems making this postponement impossible and a subsequent transition very expensive. This paper sketches an approach to specify a real-time system independent of its architecture (time-triggered or event-triggered), thus facilitating to switch between a time-triggered and an event-triggered architecture easily.