Robust Private Information Retrieval on Coded Data

TL;DR

This paper introduces robust private information retrieval schemes for coded data that can handle unresponsive nodes, ensuring optimal download efficiency regardless of node failures.

Contribution

It presents the first adaptive, universally optimal $ u$-robust PIR schemes capable of tolerating up to $ u$ unresponsive nodes in coded data settings.

Findings

Schemes are adaptive and maintain optimal download cost.

Achieve asymptotic optimality for any number of unresponsive nodes up to $ u$.

Enhance privacy and robustness in distributed data retrieval systems.

Abstract

We consider the problem of designing PIR scheme on coded data when certain nodes are unresponsive. We provide the construction of $\nu$-robust PIR schemes that can tolerate up to $\nu$ unresponsive nodes. These schemes are adaptive and universally optimal in the sense of achieving (asymptotically) optimal download cost for any number of unresponsive nodes up to $\nu$.