An automated testing experiment for layered embedded C code

This paper describes an experiment using an automated tool for testing smart cards embedded software developed in C. Conventional testing of smart cards uses low-level commands and writing the use cases and test scripts is an error-prone and tedious task. Our goal was to show how one can use a formal tool to improve the testing process in order to concentrate better on the efficiency of the test. The approach consists in modeling each layer of the system independently while abstracting the services provided by the lower layers. The model is then verified, simulated and test cases are automatically generated using test criteria such as reachable state coverage and transition coverage. We also developed a translator to execute those test cases on the C implementation of the system. We show the results obtained and the lessons learned from the application of this approach to the validation phase of a smart cards file system manager.