Automated testing of mobile applications: methodological foundations and a practical device farm solution

Abstract

Software testing activities are essential in the software development life cycle since they help to find possible bugs before releasing the product for end users. Despite its importance, testing is still not widely applied when considering the mobile ecosystem. When testing a mobile application, it is essential to consider how to execute the tests (i.e., techniques, methods, approaches, tools, and frameworks) and where the tests run (i.e., on real devices, emulators, or in the cloud). Over the years, researchers and the industry have proposed various infrastructures to facilitate test execution. However, choosing the most suitable testing infrastructure depends on the application's scope, supported features, and the company's/user's needs. Moreover, there is still a need to explore ways to optimize testing for mobile applications.Objective: This thesis aims to provide an implementation of a solution capable of running tests on multiple devices, and these tests should validate the communication between the application and an external IoT device. Moreover, the second goal is to subsidize the testing process by providing an approach for test case generation and quality assessment. In both cases, we considered two hardware components of an Android device: Bluetooth and Location.Method: This thesis presents a different methodology comprising a collection of published and submitted work from the PhD, providing distinct contributions aligned with the defined research goals. Conclusion: During this PhD, we made solid contributions by providing means to enhance mobile application testing, focusing on the Android ecosystem. In this thesis, we presented four papers published in renowned Software Engineering journals and conferences, and one paper in the submission process. We expect that the results of this work will benefit not only the academic community but also practitioners.