Refinement of Workload Models for Engine Controllers by State Space Partitioning
Morteza Mohaqeqi, Jakaria Abdullah, Pontus Ekberg, Wang Yi · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · 2017
We study an engine control application where the behavior of engine controllers depends on the engine's rotational speed. For efficient and precise timing analysis, we use the Digraph Real-Time (DRT) task model to specify the workload of control tasks where we employ optimal control theory to faithfully calculate the respective minimum inter-release times. We show how DRT models can be refined by finer grained partitioning of the state space of the engine up to a model which enables an exact timing analysis. Compared to previously proposed methods which are either unsafe or pessimistic, our work provides both abstract and tight characterizations of the corresponding workload.