MoBench: A Software Tool for Measuring Smoothness of Mobile Browsers

Mobile browser plays an important role in smartphones because it provides many valuable applications, such as data searching, online shopping, news browsing, e-mailing and so on. It is so essential that almost every smartphone is equipped with a mobile browser. One of the reasons that decides users' purchase decisions is the smoothness of human-browser interaction. Evaluating the smoothness of human-browser interaction becomes an important issue for not only mobile browser creators but also smartphone manufacturers. In order to evaluate the performance of smartphones, existing smartphone benchmarks adopted synthetic workloads, such as floating point calculation, random string sorting, memory read/write and image rendering to assess the relative performance. However, these synthetic workloads cannot fully represent the amount of work needed to be done when users interact with a mobile browser. In this paper, we present MoBench, a software tool, to directly measure the smoothness of a mobile browser. MoBench automatically triggers a smartphone under test to perform a series human-browser interactions, such as scrolling, tapping and pinching, while MoBench is still recording the interactions as a video. The video is then used to measure the response time of each interaction and determine a level of smoothness. All measurement and analysis processes of MoBench are done inside the smartphone under test without any extra hardware support. In our experiment, we first set up a series of tests to validate the accuracy of MoBench in measuring the response time of human-browser interactions. We then rank the smoothness level of four popular mobile browsers on several Android phones. Our results show that the accuracy of MoBench in measuring the response time of a user gesture is up to 99.26%. In addition, compared with existing benchmarks, MoBench can provide a more comprehensive and objective evaluation of human-browser interaction.