Template-Based Monte-Carlo Test-Suite Generation for Large and Complex Simulink Models

Takashi TOMITA  Daisuke ISHII  Toru MURAKAMI  Shigeki TAKEUCHI  Toshiaki AOKI  

IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E103-A   No.2   pp.451-461
Publication Date: 2020/02/01
Online ISSN: 1745-1337
DOI: 10.1587/transfun.2019MAP0010
Type of Manuscript: Special Section PAPER (Special Section on Mathematical Systems Science and its Applications)
model-based development,  MATLAB/Simulink,  test case generation,  Monte-Carlo method,  template-based method,  

Full Text: FreePDF(1.5MB)

MATLAB/Simulink is the de facto standard tool for the model-based development (MBD) of control software for automotive systems. A Simulink model developed in MBD for real automotive systems involves complex computation as well as tens of thousands of blocks. In this paper, we focus on decision coverage (DC), condition coverage (CC) and modified condition/decision coverage (MC/DC) criteria, and propose a Monte-Carlo test suite generation method for large and complex Simulink models. In the method, a candidate test case is generated by assigning random values to the parameters of signal templates with specific waveforms. We try to find contributable candidates in a plausible and understandable search space, specified by a set of templates. We implemented the method as a tool, and our experimental evaluation showed that the tool was able to generate test suites for industrial implementation models with higher coverages and shorter execution times than Simulink Design Verifier. Additionally, the tool includes a fast coverage measurement engine, which demonstrated better performance than Simulink Coverage in our experiments.