A hard real-time system has to produce the correct results at the intended points in time. In such a system a failure in the time domain can be as critical as a failure in the value domain. In this paper we consider the key design issues that have to be addressed in any hard real-time implementation. We then examine proposed solutions and conclude that only time-triggered architectures support an engineering approach to hard real-time system design. Finally we outline a systematic methodology for the design of a fault-tolerant time-triggered application.
展开▼