A timeliness model for time critical parallel systems

This paper deals with the problem of ensuring timeliness in safety critical real-time systems. We introduce a method to model both realtime tasks and the timeliness constraints when tasks are executed in parallel. This approach is based on a formal model for real-time tasks and on the definition of a new operator, namely the synchronized product (i.e. Pi -operator) on tasks. The Pi -operator allows us to compose the stare-transition diagrams of given tasks in order to represent their interactions together with the associated set of timeliness constraints. The Pi -operator is then used to generate a set of linear constraints, that characterizes in our approach an upper bound for the system load. This system of constraints is then used to ensure that the timeliness property can be achieved for a given real-time task set scheduled to be correctly executed in parallel. This work is done in the framework of the OASIS(1) project for the design of more reliable and safer nuclear plant control systems.