Exploiting Execution Dynamics in Timing Analysis Using Job Sequences

Abstract

Worst case design as needed for critical systems usually resorts to established methods for worst case response time analysis which rely on the worst case execution time of tasks and the minimum temporal distance between task activations. The result is often very pessimistic when compared to the real worst case load. Many feasible designs are therefore rejected under such analyses. Using worst case models based on job sequences rather than single jobs leads to less pessimistic results and makes worst case design more practical. This paper outlines existing modeling and analysis techniques which are based on job sequences and refers to several examples from automotive design where great benefits were demonstrated.

Increasing Accuracy of Timing Models: From CPA to CPA+

“…Both event and data count are important parameters to accurately quantify service request, and therefore the existing system models which consider only one of the two domains are ultimately unsatisfactory [4]. Examples for analysis frameworks with one-domain system models are Real-Time Calculus (RTC) [8], NC [5] (data/workload domain) and CPA [6] (event domain).…”