Detection of Android ad library focusing on HTTP connections and view object redraw behaviors

In recent years, the smart phone application market has expanded rapidly. One of reasons is the popularity of free applications. A developer acquires his revenues by including advertising libraries in his own application. However, some problems about these advertising libraries become clear from recent researches. Especially in the leakage of privacy information is known as a typical problem which advertising libraries cause. In order to solve this problem, the technology which detects advertisement libraries is important. In this paper, we propose a method for detection of Android ad library. We focus on the acquisition and redraw of advertising image operation which are the basic operations of mobile advertisement. Firstly, we tried running some applications with advertisements. Then, It turned out that mobile advertisements acquire advertising images from server and set that image on the screen at a fixed interval. By modifying AndroidOS, logging HTTP connections and View object redraw behaviors, we confirmed the ad image acquisition behavior. Moreover, to take advantage of the periodicity of this behavior, we carried out Fourier-transform the invocation time data of HTTP connections and redraw of View objects. Then, we extracted the periodicity by calculating correlation coefficient for these two data. From the value of correlation coefficient, it is possible to judge whether advertisement library is incorporated into an application or not. As a result, our proposal method results in a output of about 76 % detection rate.