A Reliability Evaluation Method for Probabilistic WCET Estimates Based on the Comparison of Empirical Exceedance Densities
Luís Fernando Arcaro, Karila Palma Silva, Rômulo Silva de Oliveira · 2018
Increasingly larger challenges on the determination of bounds for the Worst-Case Execution Times (WCETs) of software tasks that compose Real-Time Systems (RTSs) arise from computer architectures' performance-centric improvements. The Measurement-Based Probabilistic Timing Analysis (MBPTA) technique proposes determining probabilistic bounds for WCETs (i.e. pWCETs) by applying Extreme Value Theory (EVT) on tasks' execution time measurements, and is hence a promising approach for handling hardware complexity issues within RTSs' design. Such probabilistic WCET bounds can be associated to arbitrarily low exceedance probabilities, which are expected to be reliably respected during the system's long-term execution. In this work we propose a novel method for evaluating pWCETs' reliability, based on the comparison of empirical exceedance densities against their expected values considering target exceedance probabilities.