Schedulability and modular analysis: how to fit timing model?

In real-time systems, program execution must fulfill timing constraints and respect deadlines. Schedulability analysis checks the feasibility of a schedule according to those timing constraints. In such analysis, a task i is abstracted by a timing model, typically (at least) made of a period Ti, a deadline Di and a bound on the execution time Ci also named worst-case execution time (WCET). In real-time systems (and more generally in software engineering), the trend is to reuse code: a function is designed to be used in different contexts. For in-stance, a function that depends on the velocity of a car may accept as parameter a range from 0 to 350 km/h. This kind of information (range) may influence the execution time and thus is taken into account by the WCET analysis. How-ever, the constructor knows that the actual range depends on the vehicle: for instance, from 0 to 135 km/h for a smart ForTwo or from 0 to 315 km/h for a Nissan GTR. In this case, the estimated WCET may be largely overestimated if the velocity variable has a large influence on the executed path. For example, a large part of code may become dead code if the velocity never exceeds 200k