Research On Test Method And Key Technology Of Mobile App Testing

Mobile apps have been widely used in various domains.Today,with the increasingly fierce market competition,mobile app developers need advanced test theories and methods.However mobile app testing has some special difficults and issues,such as: thousands of mobile devices with diverse features and mobile platforms,unstable wireless network,touch screen operation,and mobility of using environment.Which make mobile apps testing more difficult and complicated.Moreover,the traditional software testing approaches are hard to meet the need of mobile app testing.Compatibility and usability are two important quality characteristics of mobile apps,which will directly influence the promotion of mobile app.Location-based service is a featured service of mobile app.In addition to conducting a systematic analysis of the compatibility,usability and location-based service of mobile apps,we also research on analyzing test risks,selecting test environments,generating and prioritizing test data,conducting automated testing and other key test techniques to improve the test efficiency and quality.The main contributions of this dissertation can be summarized as follows:(1)In terms of the compatibility testing of mobile apps,we have proposed an Apriori-based algorithm to identify compatibility features with high risks by analyzing history testing data.Meanwhile,the semantic tree test model and its coverage metrics have been proposed to drive all test environments for mobile apps.Ultimately,we propose a novel test device selection approach using multiobjective genetic algorithm.The proposed approach can recommend the minimum number of test environments to meet multiple test objects,such as test feature coverages,market occupation rates,and compatibility risks.The experiment indicates that the more compatibility bugs can be detected using the proposed approach than traditional compatibility testing approaches.(2)In terms of usability of mobile apps,a novel remote usability evaluation approach has been proposed for mobile applications.This approach automatically collects user interface events when users interact with the applications,and then models user operation tracks using state machine with time labels.By analyzing and comparing differences between user operation models with expert operation model,the effectiveness,efficiency,memorability and fault-proneness of mobile apps are evaluated and tested.The proposed approach supports to test usability in real users' environments at a large-scale.In our empirical study,the proposed approach helps to detect some usability issues about special devices and contexts,and is a good complementation for traditional usability testing.(3)For location-based information service testing,a stochastic simulation approach based on Monte Carlo sampling algorithm has been proposed to generate test data according to the distribution of point-of-interests and mobile users.The purpose of this approach achieve higher test coverage of point of interests and mobile users.And some location-related metrics are provided to select subset of test data.The experimental data shows that our method helps to improve test efficiency of locationbased information service.(4)For mobile navigation service testing,basic concepts and testing models are defined firstly.Then we propose two different methods for regression testing of mobile navigation services.One is based on the complexity of navigation zones,and the other is based on the deviation of navigation route.The complexity-based regression test approach selects and prioritizes test data by complexitizes of navigation zones.The deviation-based approach selects and prioritizes testing data depend on the deviation of navigation areas and routes.Ultimately,experiments have proved the validity of two methods,and the deviation-based approach is more effective than the complexity-based approach.