Multi-Server Weakly-Private Information Retrieval

Private information retrieval (PIR) protocols ensure that a user can download file from database without revealing any on the identity of requested to servers storing database. While existing strictly impose no is leaked file’s identity, this work initiates study tradeoffs be achieved by relaxing perfect privacy requirement. We refer such as weakly-private (WPIR) protocols. In particular, for case multiple noncolluding replicated servers, we how rate, upload cost, and access complexity improved when constraint. To quantify leakage consider mutual (MI), worst-case leakage, maximal (MaxL). present two WPIR schemes, denoted Scheme A B, based recent PIR show rate former optimized solving convex optimization problem. also achieves an compared recently proposed scheme Samy et al. under so-called $\epsilon $ -privacy metric. Additionally, family schemes partitioning presented. Moreover, provide information-theoretic converse bound maximum possible MI MaxL metrics practical restriction alphabet size queries answers. For files, tight metric, which settles capacity in particular case. Finally, compare performance their gap bound.