Private Data Aggregation over Selected Subsets of Users

Aggregator-oblivious (\(\mathsf {AO}\)) encryption allows the computation of aggregate statistics over sensitive data by an untrusted party, called aggregator. In this paper, we focus on exact aggregation, wherein the aggregator obtains the exact sum over the participants. We identify three major drawbacks for existing exact \(\mathsf {AO}\) encryption schemes—no support for dynamic groups of users, the requirement of additional trusted third parties, and the need of additional communication channels among users. We present privacy-preserving aggregation schemes that do not require any third-party or communication channels among users and are exact and dynamic. The performance of our schemes is evaluated by presenting running times.