Abstract:
A fast and responsive user interface is an essential requirement for a successful software application. Providing a responsive user interface is the challenging task of software development because there are many hurdles in identifying the responsiveness of the application. Methods used in usability inspection are still non-automated procedures that require complex user testing procedures or lengthy questionnaire from user studies. The results obtained from these procedures are not very helpful and informative because each user expresses their view of the convenient interface, which can conflict with other user’s preferences. Another problem with these methods is that the prototype or real application interface is required for its testing. Alternatively, few studies have also proposed automated evaluation methods that require software and tools to collect data, and very limited cases of such tools are available. In most of the cases, data analysis is still a manual process. However, this study proposes an automated analysis mechanism called Latency Based Interval Model (LIBM) for measuring the responsiveness of a Graphical User Interface (GUI) application. A simulation technique is used to create a large number of input events on behalf of the user that facilitates to capture a substantial amount of log data from the application and correlates it to determine the weak points. To improve the responsiveness, LIBM calculates the total time that the application takes to process the user-generated events. The study also measures various types of delays introduced by hardware and platform jitter to exclude it from the application delay for proper reflection of accurate measures. The entire exploratory study is executed utilizing an arrangement of scripts to perform computerized testing of the applications without the user intervention to drive its functions. Thesis study has carried out a large number of trials on different Java and Android applications including a case study on Circuit Recognizer (Voltique Designer) application that highlighted many performance and accuracy issues. On successful implementations of ParaTask parallelism in corresponding applications, Android applications specifically shown significant improvements as compared to their sequential versions, in terms of faster response times and smaller variances.