| KEYWORD |
Comparison of model-based testing tools for Android apps
keywords ANDROID, SOFTWARE TESTING
Reference persons LUCA ARDITO, MAURIZIO MORISIO
External reference persons COPPOLA RICCARDO
Research Groups DAUIN - GR-16 - SOFTWARE ENGINEERING GROUP - SOFTENG
Thesis type EXPERIMENTAL
Description Android app testing can be performed by means of several different techniques, e.g., manually-written scripts to identify interactable GUI elements, the registration of real input sequences on the GUI, or the adoption of image-recognition based and model-based techniques that allow the generation (either scripted or scriptless) of test cases to exercise the application GUI.
Many tools have been proposed recently in the field of model-based testing. These tools are based on an abstract representation of the GUI (graphical user interface) of the application, e.g. in the form of graphs, or finite state machines. These representations are used for the automated generation of test sequences, and can themselves be automatically generated by the tool or manually defined by the testers. Most recent techniques allow the automated generation of verifications to be executed against the running application, to verify that it is behaving properly. The model-based approach reduces significantly the effort needed to create test suites, and a higher coverage of widgets and pages in the application. The main drawback is a possible difference between the generated sequences, and realistic user interactions with the application.
The objective of this master's thesis is a thorough analysis of the most recent solutions for model-based testing of mobile applications; the analysis will include the definition of a set of metrics to perform a comparison of the tools, and the execution of an empirical experiment to measure these metrics on a set of real-world mobile projects.
Required skills Java, basi di software testing, basi di programmazione Android
Deadline 28/10/2021
PROPONI LA TUA CANDIDATURA