The main focus of this paper is the problem of ensuring timeliness in safety critical systems. First, we introduce a method and its associated technique to model both real-time tasks and the timeliness ensuring concern when tasks are executed in parallel. This approach is based on formal aspects of our real-time tasks model and on the definition of the synchronized product operator on the tasks. Real-time tasks are equivalent to their state-transition diagrams and the operator allows us to compose the diagrams of a set of tasks to represent their interactions. The operator is then used to map the tasks to a system of linear constraints to determine the schedulability of the tasks and deduce a system load upper bound. An illustration of our technique on a safety critical study case is presented in which the timeliness property can be achieved for the real-time set of tasks executed in parallel on the same processor. We also introduce how this work can be applied to the multiprocessor case.