Model-Based Testing of Software-Based System Functions

Abstract

Software is the most crucial innovation driver for cyber-physical systems and validating its correct behavior is crucial to many application domains from automated vehicles to medical systems, smart homes, and smart cities. Typical for these systems is that software errors can cause severe ramifications and can be very costly to fix. Consequently, facilitating a priori software validation and verification is of crucial importance for cyber-physical system development. We hence present a modelbased method which facilitates test creation, test maintenance, and improved test coverage in automotive software engineering. To this end, we translate specification models used in the BMW Group's SMArDT engineering methodology into activity diagrams, apply optimizations, and ultimately generate test cases. We realize these transformations using the MontiCore language workbench and report on ongoing evaluation at the BMW Group. First results indicate that model-based test case creation can provide great benefits to all stakeholders, once the initial learning efforts have been mastered.