Handling sporadic tasks in off-line scheduled distributed real-time systems

Many industrial applications mandate the use of a time-triggered paradigm and consequently the use of off-line scheduling for reasons such as predictability, certification, cost, or product reuse. The construction of an off-line schedule requires complete knowledge about all temporal aspects of the application. The acquisition of this information may involve unacceptable cost or be impossible. Often, only partial information is available from the controlled environment. In this paper we present an algorithm to handle event-triggered sporadic tasks, i.e., with unknown arrival times, but known maximum arrival frequencies, in the context of distributed, off-line scheduled systems. Sporadic tasks are guaranteed during design time, allowing rescheduling or redesign in the failure case. At run-time, the sporadic tasks are scheduled dynamically, allowing the reuse of resources reserved for, but not consumed by the sporadic tasks. We provide an off-line schedulability test for sporadic tasks and apply the method to perform on-line scheduling on top of off-line schedules. Since the major part of preparations is performed off-line, the involved on-line mechanisms are simple. The on-line reuse of resources allows for high resource utilization.