DeSeCT: uma estratégia heurística usando atributos de aplicações móveis na seleção de dispositivos para testes de compatibilidade

Abstract

Ensuring the quality of mobile applications and achieving greater device market coverage is increasingly necessary, as there is a diversity of mobile devices, brands and operating system versions. The execution of compatibility tests is preceded by the selection of devices on which the application will be tested. Selecting devices on which the tests will be performed is an important task, but deciding how many and which ones will be used for the test execution, in the face of thousands of device models, is not a trivial task. Existing approaches are mainly based on device characteristics, thus disregarding the characteristics of the application being tested. Therefore, in this work, DeSeCT (Devices Selection for Compatibility Testing) is proposed. This proposal is based on the NSGA-II algorithm and uses device information and attributes of the application being tested in order to minimize the number and cost of selected devices and maximize the coverage of characteristics such as screen size, resolution, network, version api and marketshare. A device dataset was also built with 4039 mobile devices from brands worldwide. The results of the first study with 30 applications made it possible to identify the importance of using the attributes of the application, which allowed us to improve the search space and have solutions with fewer devices and high coverage of device characteristics. To identify the state-of-practice, a Survey was carried out, which allowed us to identify the three criteria most used by industry professionals in the device selection process, which are: marketshare, screen size, and OS version. Other factors that impact the decision are location and known brands. DeSeCT presents a set of solutions with the relationship between the cost of the solution and its coverage of device features, aiming to assist in decision-making in the mobile device selection process.