Android side channel attacks

Abstract

In the last decade, the smartphone boom made it accessible to everyone and everywhere. These days almost everyone has a smartphone and among them, most of the people spend most of their time on the smartphones. And smartphones handle many of sensitive user information, plus the things someone does on their smartphone also tells a lot about their behavior. And an individual’s behavior and routine of using the phone can often become highly valuable to interested malicious actors. In this study, we investigate the potential of side-channel attacks as a method to learn user’s behavior from their Android devices. By analyzing existing side-channel attack techniques, through system call signals, we develop a novel approach that leverages system metrics to infer user behavior and app interactions. We devised a way to collect system calls from several phones during app interactions then trained and tested a CNN-GRU model to predict app-task pairs from system call signals.