A Method for Determining Whether a Simulink Model Is Ready for Test Generation

EasyChair Preprint 4532

Abstract

Model-based development (MBD), in which development specifications are written in Simulink, is widely used in the development of control system embedded systems. We need to create test cases to verify Simulink models. Automated test generation tools are used to reduce the effort to develop test cases. However, depending on how the model is written, the Automated generation tool may fail, and it takes a long time to determine generation failure. This paper proposes a method for predicting the feasibility of generating test cases for the model. In particular, we evaluate the validity of feature generation using the Bag of nodes representation of the method and the summary statistics of the graphs. The results show that the AUC of 0.628 is not quite accurate enough to be practicable, but the initial results using a large amount of data are promising.

Keyphrases: Simulink, directed graph, machine learning, model-based development

@booklet{EasyChair:4532,
  author    = {Takuya Ogata and Yuge Liu and Kenji Hisazumi},
  title     = {A Method for Determining Whether a Simulink Model Is Ready for Test Generation},
  howpublished = {EasyChair Preprint 4532},
  year      = {EasyChair, 2020}}