Mobile apps often use analytics infrastructures to gather data on app performance and user behaviors. There are significant privacy concerns about the collection and use of such data. Legislative efforts and societal demands are calling for increased transparency and well-defined compromises between the utility of data gathering and the corresponding loss of privacy. This project will improve the privacy of mobile app analytics by employing differential privacy, which provides a rigorous theory and a powerful algorithmic framework for privacy-preserving analysis. Integration of research and education will develop the expertise of developers of mobile apps and data analyses. Recruitment of underrepresented students will contribute to increased diversity in computing. The project will advance the important area of user privacy, in an environment with rapidly-increasing societal and legislative demands for privacy.
The investigators will develop rigorous and quantifiable privacy mechanisms in mobile-app analytics by leveraging the theory of local differential privacy (LDP). The first contribution of the project will be the design and implementation of PrivAid, a conceptual approach and related software tools for deploying LDP analyses in widely-used infrastructures for mobile-app analytics. Based on this foundation, several algorithms targeting important software analyses will be developed using novel techniques for randomization and sampling: (1) edge/path profiling in control-flow graphs; (2) analyses of heavy hitters for event profiling and graph profiling; (3) profiling under continuous observation. The public release of PrivAid, software analyses, and experimental subjects will enable the development and evaluation of future LDP software analyses.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
