Resource sharing of real-time applications with other applications

Abstract

Most real-time applications run on dedicated system, to eliminate the uncertainty that may arise if other applications are running on the system. The resource requirement for the real-time application is determined on the basis of the system's ability to provide fault tolerance under the worst case execution time assumption. Costs of such systems will be better justified, if the real-time applications share the resources with other non real-time applications (such as, batch jobs), without compromising the real-time deadlines. Assuming a real-time application to consist of a set of optional and mandatory subtasks with time-critical precedence constraints among them, the problem of scheduling these subtasks on such a shared system so as to optimize the use of resources and execute time-critical applications with sufficient accuracy is addressed.<>