Efficiently Testing AUTOSAR Software Based on an Automatically Generated Knowledge Base

The system architecture AUTOSAR divides an automotive ECU into three main layers—application layer, RTE and basic software layer. Due to clearly specified interfaces, each of these layers and its software modules can be developed partly or completely by different companies with different tool chains. This promotes competition between companies and enables the hardware independence of applications, but it also leads to challenges during the integration phase of an ECU, the mapping towards different ECUs, and the reuse of functions. To ensure a high quality of AUTOSAR developments, we provide an approach that checks automatically the source code and configuration files of a project before compilation, i.e., performing static tests. Our approach is independent of the tool chain used and of the AUTOSAR version. The consistency and AUTOSAR compliance tests are based on a knowledge base which is filled semi-automatically by our parsers. The knowledge base contains information about the structure of an AUTOSAR ECU, i.e., basic software modules, layers, function names and parameters, hardware information, etc. We create the information by parsing once AUTOSAR meta-models, base projects and configuration files of the tools automatically. These inputs provide most of the information, necessary for our tests. Our static tests check the consistency between configuration files and source files, validate the configuration, and parse source files for non-compliant code within the given AUTOSAR project. Besides a test report, we visualize test results in a graphical user interface (GUI) which shows the layers and modules of the AUTOSAR project. Data for visualization is further provided by our knowledge base. All problems like inconsistencies or missing modules are visualized by flags on the corresponding elements to support testers and developers finding solutions. Our approach increases the quality of AUTOSAR projects by detecting inconsistencies, errors, and non-compliant code with AUTOSAR. In addition, by finding inconsistencies/errors in an early phase, our semi-automated test approach provides an optimal basis for the dynamic test of an AUTOSAR ECU supporting and complementing acceptance tests. Finally, all known AUTOSAR tool chains with readable configuration files and source files can be integrated with our approach.