FrauDroid: An Accurate and Scalable Approach to Automated Mobile Ad Fraud Detection

Previous studies on mobile ad fraud detection are limited to detecting certain types of ad fraud (e.g., placement fraud and click fraud). Dynamic interactive ad fraud has never been explored by previous work. In this paper, we propose an explorative study of ad fraud in Android apps. We first created a taxonomy of existing mobile ad fraud. Besides static placement fraud, we also created a new category called dynamic interactive fraud and found three new types of them. Then we propose FrauDroid, a new approach to detect ad fraud based on user interface (UI) state transition graph and networking traffic. To achieve accuracy and scalability, we use an optimized exploration strategy to traverse the UI states that contain ad views, and design a set of heuristic rules to identify kinds of ad fraud. Experiment results show that FrauDroid could achieve a precision of 80.9% and it is capable of detecting all the 7 types of mobile ad fraud. We then apply FrauDroid to more than 80,000 apps from 21 app markets to study the prevalene of ad fraud. We find that 0.29% of apps that use ad libraries are identified containing frauds, with 23 ad networks suffering from ad fraud. Our finding also suggests that Dynamic interactive fraud is more prevalent than static placement fraud in real world apps.