An analysis of mobile application network behavior

Abstract

The impact of mobile applications in our every day lives is indisputable. While there are millions of applications available in mobile application markets, users often have very little information to rely upon when selecting which applications to download and use in terms behavior and security. Many applications may have the functionality users desire but may also have undesirable behavior such as excessive network usage, use of ads, linkage to malicious sites, or unauthorized leakage of user or application usage information. In this paper, we design a system to provide users with additional information on the application's behavior by actively eliciting and identifying undesirable behavior exhibited by mobile applications. In order to demonstrate the feasibility of our system, we use it to study the network behavior of top mobile applications listed in the Android and iOS application markets. We find that applications can be clustered into groups that have distinct behavior patterns. Using our system, users can determine an application's behavior group prior to installation on their mobile devices, reducing the exposure to any undesirable application behavior.