Estimating probabilistic timing performance for real-time embeddedsystems

摘要

In system-level design of real-time embedded systems, being ablento capture the interactions among the tasks with respect to timingnconstraints and determine the overall system timing performance is anmajor challenge. Most previous works in the area are either based on anfixed execution time model or are only concerned with the probabilisticntiming behavior of each individual task. The few papers that deal withnoverall system probabilistic behavior have used improper assumptions. Innthis paper, given that the execution time of each task is a discretenrandom variable, a novel concept of state is introduced based on a newnmetric that is derived that measures the probability of a task set beingnable to be scheduled. Several approaches to evaluating the metric arenalso presented. Applying this metric in the system-level designnexploration process, one can readily compare the probabilistic timingnperformance of alternative designs