Abstract
Thorough testing of software is necessary to assure the quality of a product before it is released. The testing process requires substantial resources in software development. Model-based software development provides new possibilities to automate parts of the testing process. By automating tests, valuable time can be saved.
This thesis focuses on different ways to utilize models for automatic generation of test vectors and how test coverage analysis can be used to assure the quality of a test suite or to find “dead code” in a model. Different test-automation techniques have been investigated and applied to a model of an adaptive cruise control system (ACC) used at Scania. Source code has been generated automatically from the model, model coverage and code coverage has therefore been compared.
The work with this thesis resulted in a new method to create test vectors for models based on a combinatorial test technique. Model-based development and coverage analysis open new doors to automate testing. In this thesis,automated testing of software developed in a model-based environment such as Simulink has proven tobe both more time-efficient and more accurate than previously used manual tests. The use of a model-coverage tool such as Simulink Verification & Validation provides possibilities to verify the accuracy ofmanual tests as well as new possibilities to automate a substantial part of the testing process.
Source: Linköping University
Author: Andersson, Jonny