PUBA: Privacy-Preserving User-Data Bookkeeping and Analytics

Abstract In this paper we propose Privacy-preserving User-data Bookkeeping & Analytics (PUBA), a building block destined to enable the implementation of business models (e.g., targeted advertising) and regulations fraud detection) requiring user-data analysis in privacy-preserving way. PUBA, users keep an unlinkable but authenticated cryptographic logbook containing their historic data on device. This can only be updated by operator while its content is not revealed. Users take part analytics computation, where it ensured that up-to-date authentic potentially secret function verified privacy-friendly. Taking constrained devices into account, may also outsource analytic computations (to malicious proxy colluding with operator).We model our novel Universal Composability framework provide practical protocol instantiation. To demonstrate flexibility sketch instantiations detection advertising, although could used many more scenarios, e.g. for multi-modal transportation systems. We implemented bookkeeping protocols exemplary outsourced computation based logistic regression using MP-SPDZ MPC framework. Performance evaluations smartphone as user device powerful hardware suggest PUBA smaller logbooks indeed practical.