Automated classifier generation for application-level mobile traffic identification

This paper proposes an automated classifier generation system for application-level mobile traffic identification. The proposed system comprises traffic classifier generation system architecture and mobile traffic measurement agents (mTMAs) that are installed on mobile devices to monitor their application-level traffic. To generate classifiers, our adaptive algorithm selects the lowest cost classifiers from among the available classifier candidates, while assuring acceptable identification accuracy. To verify the efficacy of the proposed identification system, we implemented and deployed it on a campus network. The results of mobile application traffic identification obtained using the system in this deployment are also presented in this paper.