Co-Evolution of Model-Based Tests for Industrial Automotive Software

Abstract

Model-based software is evolving at an increasing rate, and this has an impact on model-based test suites, often causing unnecessary regeneration of tests. Our work proposes that by examining evolution patterns of Simulink automotive models and their associated test models we can identify the direct impacts of evolution on the tests. Using these evolution patterns, we propose the design of a process to ensure that as a Simulink model evolves its associated test models are automatically adapted, requiring minimal computation. This will lead to the development of a prototype tool capable of performing this model- based test co-evolution of tests alongside source models and presenting results to test engineers.