The Extreme Value Theory in Probabilistic Real-Time Computing

In this article we introduce the statistical theory foundation of probabilistic real-time computing: the Extreme Value Theory. This tutorial aims at introducing the basis of the theory and provide some insights on how to apply it to the real-time computing problems. In fact, a method called Probabilistic Real-Time Computing has been recently proposed to overcome the issues in estimating the worst-case execution time with traditional approaches on modern processing architectures. Provided that certain hypotheses are satisfied, the Extreme Value Theory is able to produce a probabilistic estimation of the worst-case timing behaviour of a real-time task, i.e. the statistical distribution of its execution times.