Critérios de geração de casos de teste de sistemas de tempo real.

Abstract

Nowadays, computer systems have increasingly taken place in society in many different sectors. You can find software operating systems in patient monitoring systems, air traffic control, robotic systems, vehicles, etc.. The majority part of this systems work under time constraints. Systems whose functioning depends not only on the correctness of the outputs produced, but also on the moment when these outputs are generated are known as real-time systems. Testing is an activity that demands a very high cost, but testing real-time systems is even more challenging and costly. Thus, Model Based Test has been a widely used technique to generate test cases for both systems in general and for real-time systems. So, model based test cases generation tools receive as input the model of the system under test and the way at which the tests will be generated. The way this tests are generated is known as generation criterion. The generation criterion reflects the test case generation algorithm and, therefore, determines which model elements will be covered. There are other works that study criteria for generating real-time systems, but do not analyze them watching their effectiveness, ie. relating the size of the set of test cases generated with ability to reveal faults. It were not find in the literature works that examine generation criteria for real-time systems in this sense. Thus, it was proposed in this work a set of generation criteria to generation of real-time systems model based test cases identified through a systematic review. For this work was chosen the symbolic model HOSTS to describe real-time systems. The SYMBOLRT tool was extended in order to support the generation of test cases based on generation criteria and an experimental study was performed using six different models and executed along with all the selected generation criteria. By the experimental study, we concluded that there is difference between the investigated criteria in relation to the ability to reveal faults and the size of the set of test cases generated by each generation criterion. Thus, some subjective conclusions could be made which can be used to assist the tester when choosing which criteria to adopt for generating test cases for real-time systems.