As the technology in cars continues to evolve, the development is faced with increasingly complex software solutions. Tasks such as maintenance over the lifespan of the vehicle, reuse in other models as well as compliance with regulatory requirements, therefore, need to be efficiently accomplished to keep up with ever shortening development cycles. In the context of the research project SmartDelta (sponsored by the Federal Ministry of Education and Research on the basis of a decision by the German Bundestag via ITEA4), we are developing a tool for code-based similarity analysis. This enables us to obtain automated code reuse suggestions, which makes code changes such as fixing vulnerabilities over multiple product variants faster. Combined with test case prioritization, it will significantly decrease time needed for testing. By taking advantage of the collaboration possibilities within the project, our solution is created with large, production scale codebases in mind. Our approach is based on Code Property Graphs which provide a compact yet comprehensive model on the code. We employ classical algorithms in conjunction with machine learning to extract the desired information out of Code Property Graphs.
Session:
Poster
|