MISE 19 - Montreal, Canada/Posters

From MiSE

Jump to: navigation, search


On the Characterization of Model Families and Union Models
Sanaa Alwidian, Daniel Amyot

A model family is a set of related models, in a given language, with commonalities and variabilities between individual models. Model families result from the evolution of models over time (i.e., versions) and/or variation over the space dimension, such as in (software) product lines. In situations where there is a large number of models in a family, it may become more efficient to analyze and reason about models, all at once, instead of analyzing each model separately. To achieve this, we propose the concept of union model as a first-class generic artifact that captures a model family (in both dimensions of variability) in a compact and exact manner, to enable efficient analysis. This work provides a characterization of “model families” and “union models”, together with a simple graph-based formalization. It also presents how variability among members of a model family can be distinguished in union models using a spatio-temporal annotation language. We define several experiments to analyze the performance gains of using union models and conducted one of these experiments. Our preliminary empirical results show that the use of union models (representing multiple models) has the potential to increase analysis efficiency compared to the analysis of individual models.

Model Driven Generation of an Overall Scheduling in Cooperative Component Based Development
Arthur Strasser, Christoph Knieke, Andreas Rausch

In the automotive domain the development of embedded real-time systems is organized by a cooperative development between the OEM (Original Equipment Manufacturer) and a set of suppliers. The OEM as the vehicle manufacturer is responsible for system development and integration whereas subsystems and essential parts of the embedded software components are developed by suppliers. The integration task however comes with difficulties: Following the AUTOSAR standard for system development, the OEM is not able to specify scheduling requirements for an overall scheduling as the composition into software components – which is the basis for an overall scheduling – takes places in the subsequent subsystem design by the suppliers. Thus, during system integration the OEM has to determine the overall scheduling with expert knowledge and by testing possible scheduling options at great expense. To facilitate the generation of an overall scheduling, this work proposes a methodology based on an extension of the AUTOSAR standard enabling the specification of requirements for an overall scheduling before the decomposition into software components takes place. If a correct overall scheduling exists, this scheduling can be generated based on the previously specified scheduling requirements.

  • This page was last modified on 20 March 2019, at 14:16.
  • This page has been accessed 346 times.