Section: New Results

Probabilistic and statistical temporal analysis

Participants : Liliana Cucu, Robert Davis, Adriana Gogonel, Walid Talaboulma, Dorin Maxim, Cristian Maxim.

Real-time constraint guarantees require worst-case reasoning to provide sound solutions. We have proposed to define and use worst-case reasoning in different contexts: optimal scheduling algorithms, response time analysis, estimation of worst-case execution times. These results have laid the foundations for certifiable probabilistic solutions to real-time systems.

In particular, we have studied the probabilistic response time analysis for systems with multiple probabilistic parameters, either by using bounds based on real-time calculus, extreme value theory, direct calculation or in a context of component-based systems. Generally, probabilistic methods have high complexity cost; using upper-bounds for the input probability distributions we provide conservative(safe) results faster. Worst-case reasoning is also provided for the statical estimation of a task probabilistic worst-case execution time.

Results were published in [22] , [24] , [58] , [56] , [42] , [46] , [23] , [42] , [43] , [40]