Summary: | 碩士 === 元智大學 === 資訊工程學系 === 107 === Because there exists a severe fragmentation problem for Android devices,program developers need to consider various screen resolutionsin GUI layout design. They may spend a lot of time in inspecting the GUI layouts to detect layout anomalies. Recently, several tools for layout anomaly detection have been proposed. However, they are still time-consuming because of their sequential testing nature. In this study, a distributed layout testing framework, DiLAD, is proposed to improve the testing performance of layout anomaly detection. DiLAD employs the longest-processing-time-first (LPT) scheduling algorithm to dispatch the detection tasks. Through the distributed layout anomaly detection architecture, DiLAD tries to evenly distribute the detection tasks to testing slaves from the master. Therefore, the testing performance can be improved. We have conducted empirical experiments to study the effectiveness of DiLAD. The experimental results show that the distributed layout testing framework can effectively improve the testing performance. In addition, the employed LPT scheduling algorithm outperforms the RR algorithm when DiLAD has more than one slave.
|