Full Program »
Adaptation oriented test data generation for Adaptive Systems
The complexity of modern and emerging software systems has steadily increased. These systems must adapt to preserve their operation in the presence of uncertain changes. One of the techniques of quality assurance is software testing. The generation of test data sets presents a great challenge for the test activity due to the multiple inputs, combinations of variables and dependence on the environmental information in which the adaptive system is inserted. This research presents an approach, called AGAS, for the generation of test data sets for adaptive systems. AGAS uses the concept of virtual environments, the changes in these environments and the implications of system adaptations and works with the interaction of the control loop of an adaptive system. An experimental study was carried out to compare AGAS with the randomized approach using three assessment metrics. The approach was applied to three examples of adaptive systems. The results obtained showed that there was a considerable gain in the generation of the test data set, occurrences of adaptations and adaptations with errors. AGAS suggest that it is a promising approach to generate test data sets for adaptive systems considering the adaptation criterion.