Project Details
Projekt Print View

CPOT-SM – Complete Property-oriented Testing with Symbolic Methods

Subject Area Software Engineering and Programming Languages
Security and Dependability, Operating-, Communication- and Distributed Systems
Theoretical Computer Science
Term from 2019 to 2024
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 407708394
 
Final Report Year 2024

Final Report Abstract

The main objective of this project was to provide complete testing methods (these methods provide correctness guarantees for implementations passing the tests) to verify that implementations under test (IuT) fulfil their specified properties (so-called property-oriented testing (POT)). Automating POT and, at the same time, ensuring completeness guarantees is of considerable importance for the verification of safety-critical systems, where the strength of the testing methods applied needs to be justified, and test campaigns with conventional means become unacceptably costly. The complexity problem often limiting the practical applicability of complete test suites has been mitigated in this project by the creation of symbolic system representations based on several novel types of equivalence classes. With the help of these classes, the behavioural model of an IuT can be abstracted to a manageable size, but without losing the capability to uncover all violations of the specified properties. For two complementary POT approaches, novel complete methods have been developed, each of them able to cope with realistic IuT complexity, as found in airbag controllers, speed monitors, or train protection systems. The first approach is applicable for developments following the model-based systems and software engineering paradigm, where design models are created in addition to the required properties. The existence of a model that is already known to fulfil the required property allows to ensure property fulfilment with significantly fewer test cases. The second approach is specialised on software module testing and applies machine learning techniques for automated internal model creation and subsequent model checking, so that test engineers only have to specify the required properties. An open source software ecosystem has been created implementing both POT approaches. The test generation and execution software is runnable in a cloud server farm and can be accessed freely for research and evaluation purposes.

Publications

 
 

Additional Information

Textvergrößerung und Kontrastanpassung