Maximal software execution time: a regression-based approach

This work aims at facilitating the schedulability analysis of non-critical systems, in particular those that have soft real-time constraints, where worst-case execution times (WCETs) can be replaced by less stringent probabilistic bounds, which we call maximal execution times (METs). To this end, it is possible to obtain adequate probabilistic execution time models by separating the non-random dependency on input data from a modeling error that is purely random. The proposed approach first utilizes execution time multivariate measurements for building a multiple regression model and then uses the theory related to confidence bounds of coefficients, in order to estimate the upper bound of execution time. Although certainly our method cannot directly achieve extreme probability levels that are usually expected for WCETs, it is an attractive alternative for MET analysis, since it can arguably guarantee safe probabilistic bounds. The method’s effectiveness is demonstrated on a JPEG decoder running on an industrial SPARC V8 processor.