Mobile applications play an important role in the dissemination of computing and information resources. They are often used in domains such as mobile banking, e-commerce, and health monitoring. Cost-effective testing techniques in these domains are critical. This dissertation contributes novel techniques for automatic construction of mobile application test suites. In particular, this work provides solutions that focus on the prohibitively large number of possible event sequences that must be sampled in GUI-based mobile applications. This work makes three major contributions: (1) an automated GUI testing tool, Autodroid, that implements a novel online approach to automatic construction of Android application test suites (2) probabilistic and combinatorial-based algorithms that systematically sample the input space of Android applications to generate test suites with GUI/context events and (3) empirical studies to evaluate the cost-effectiveness of our techniques on real-world Android applications. Our experiments show that our techniques achieve better code coverage and event coverage compared to random test generation. We demonstrate that our techniques are useful for automatic construction of Android application test suites in the absence of source code and preexisting abstract models of an Application Under Test (AUT). The insights derived from our empirical studies provide guidance to researchers and practitioners involved in the development of automated GUI testing tools for Android applications.

Mobile applications are growing in popularity and pose new problems in the area of software testing. In particular, mobile applications heavily depend upon user interactions and a dynamically changing environment of system events. In this thesis, we focus on user-driven events and use Q-learning, a reinforcement machine learning algorithm, to generate tests for Android applications under test (AUT). We implement a framework that automates the generation of GUI test cases by using our Q-learning approach and compare it to a uniform random (UR) implementation. A novel feature of our approach is that we generate user-driven event sequences through the GUI, without the source code or the model of the AUT. Hence, considerable amount of cost and time are saved by avoiding the need for model generation for generating the tests. Our results show that the systematic path exploration used by Q-learning results in higher average code coverage in comparison to the uniform random approach.