Implicit Testing: Rethinking Efficiency, Quality, and Teamwork

Traditionally, embedded software in automotive control units is validated using an explicit testing approach; requirements are manually translated into test cases, often resulting in high maintenance effort and limited effectiveness. However, in practice, explicit testing has shown major drawbacks. It’s hard to scale, rarely consistent, and often creates friction between development and testing teams.

With implicit testing, we introduce a paradigm shift: instead of validating each requirement in isolation, the system operates within realistic, automatically generated scenarios that are evaluated against high-level validation criteria. This creates a system-level, test-driven run-time environment where bugs are revealed without needing to be explicitly defined.

The testing setup is split into three independent components:

• Plant models developed in Simulink® and Simscape™ simulate realistic physical and hard-ware behavior.
• Test scenarios representing customer use cases and fault conditions, which are generated automatically using simple keywords or recorded MF4 data ingested via Vehicle Network Toolbox™.
• Validation criteria implemented in MATLAB^® to assess overall system behavior based on high-level requirements, independent of software implementation details.

We’ve successfully implemented this approach with these outcomes:

• Drastically reduced test maintenance effort with significant quality improvements.
• Early detection of unexpected bugs through system-level testing.
• Improved collaboration with clear responsibilities and traceable root causes.
• Scalable test coverage through reusable scenarios and criteria.

Implicit testing offers an efficient and scalable approach to validating complex embedded systems. This presentation outlines the methodology, application, and impact in our projects.