Model Checking of Automotive Control Software: An Industrial Approach

Masahiro MATSUBARA  Tatsuhiro TSUCHIYA  

Publication
IEICE TRANSACTIONS on Information and Systems   Vol.E103-D   No.8   pp.1794-1805
Publication Date: 2020/08/01
Online ISSN: 1745-1361
DOI: 10.1587/transinf.2019FOP0002
Type of Manuscript: Special Section PAPER (Special Section on Formal Approaches)
Category: Formal Approaches
Keyword: 
model checking,  program slicing,  automotive control systems,  SPIN model checker,  

Full Text: PDF(1016.6KB)>>
Buy this Article




Summary: 
In automotive control systems, the potential risks of software defects have been increasing due to growing software complexity driven by advances in electric-electronic control. Some kind of defects such as race conditions can rarely be detected by testing or simulations because these defects manifest themselves only in some rare executions. Model checking, which employs an exhaustive state-space exploration, is effective for detecting such defects. This paper reports our approach to applying model checking techniques to real-world automotive control programs. It is impossible to directly model check such programs because of their large size and high complexity; thus, it is necessary to derive, from the program under verification, a model that is amenable to model checking. Our approach uses the SPIN model checker as well as in-house tools that facilitate this process. One of the key features implemented in these tools is boundary-adjustable program slicing, which allows the user to specify and extract part of the source code that is relevant to the verification problem of interest. The conversion from extracted code into Promela, SPIN's input language, is performed using one of the tools in a semi-automatic manner. This approach has been used for several years in practice and found to be useful even when the code size of the software exceeds 400 KLOC.