A Methodology for the Simplification of Tabular Designs in Model-Based Development

摘要

Model-based development (MBD) is increasingly being used to develop embedded control software, with Matlab Simulink/Stateflow being the most widely used MBD language in the automotive industry. Stateflow truth tables, more traditionally known as decision tables, are often used for implementing complex decision-making logic. As the subsystems utilizing State flow truth tables evolve, they often grow more complex and become difficult to maintain and test. It is in part due to the nature of decision tables that makes them more difficult to check for desirable properties such as disjoint ness and completeness, resulting in reduced readability and scalability. Tabular expressions provide an alternative representation which does not suffer from many of the same problems. With the safety-critical nature of the automotive domain, as well as the continuous growth in both size and complexity of models, well-defined and principled methodologies are required for maintaining and refactoring tables. This paper presents a refactoring methodology for simplifying decision tables through the use of tabular expressions to facilitate testing, traceability and readability to help companies comply with ISO 26262. An automotive industrial case study is used to motivate the work and demonstrate the proposed methodology.