Efficient Exploration of Complex Data Flow Models

Abstract

The modeling tools that are commonly used for embedded software development are rather limited when it comes to communicating certain model properties between different groups of engineers. For example, calibration engineers need to understand dependencies between signals and calibration parameters, while function developers create models with a divide-and-conquer strategy, where details of signal dependencies are hidden by abstract interfaces. We state requirements for modeling tools to improve the exploring of complex data flow models and to facilitate the understanding of engineers from different domains. We propose an approach that combines transient views and automatic layout and present two implementations based on different technologies, GMF and KLighD. While both technologies fulfill all requirements, KLighD turned out to be superior in terms of both performance and programming effort. The implementations are based on an open-source framework and are employed in a commercial product that targets the calibration process for automotive software development.