Iraklis Leontiadis - PhD Student, RS department Digital Security
Date: - Location: Eurecom
Owing to the widespread deployment of ubiquitous devices, end users are burgeoning service providers with massive amount of data. This proliferation of information has enabled new paradigms in data collection and processing. However this new paradigm acts as a double-edged sword: on the one hand the collection of information empowers service providers to compute useful aggregate statistics, but on the other hand serious privacy concerns are undermined. In this presentation we will focus on privacy preserving computations performed by an untrusted aggregator on highly sensitive data whereby the untrusted aggregators are granted to learn the result of the computation, thus relaxing standard cryptographic data confidentiality definitions. We will first review current state of the art work for privacy preserving computations in the multi-user setting and we will describe a novel solution tailored for dynamic populations with a more powerful threat model.