EAGER: A New Framework for Mobile Network Monitoring, Learning and Control

University Of California-Irvine, Irvine CA

Investigators

Abstract

Mobile devices generate an ever-increasing volume of traffic, are used for a range of applications from communication to financial transactions, and have access to personal information. Since mobile user behavior as well as third-party activities eventually manifest themselves through using the network, passive network monitoring offers a unique opportunity to detect both legitimate and malicious activity patterns on the mobile device. This project proposes AntMonitor - a new framework for real-time, on-device, passive network monitoring and crowd-sourcing. The goal is to understand, learn and control patterns in network activity, for applications related to privacy, security, performance, and behavioral analysis. The research agenda promotes transparency of mobile data, puts the user in control of how her data are shared or monetized, and can inform policy makers. Given today's size and personal nature of mobile data, changing the practices of how mobile devices handle and share our information can have significant societal impact, primarily in terms of privacy and security and secondarily in terms of the economics of personal data. In addition, the project will train students and minorities, and will provide software tools and data sets to the research community. This project will build and deploy AntMonitor - a system for collection and analysis of fine-grained, large-scale, passive network measurements from mobile devices. Design challenges that will be addressed include high performance in terms of network throughput and battery consumption, and modular design so as to support different applications including (i) privacy leaks detection and prevention (ii) learning of user and app behavior and anomaly detection and (iii) network performance monitoring. Each of these application domains requires its own module in the AntMonitor framework and faces its own challenges in terms of system design, algorithms and data analysis. Overall, the project will advance the state-of-the-art in mobile network monitoring and will improve our understanding of patterns in mobile network activity. It will produce novel algorithms and data analysis methods that enhance the performance, security and privacy of mobile devices. A unique challenge lies in crowd-sourcing and deploying AntMonitor with real users in the wild. To this end, the project will explore different ways to popularize the technology, including user-facing apps, libraries, open-source software, and data-sets available to the community.