Quantum $X$-Secure $B$-Byzantine $T$-Colluding Private Information Retrieval
Mohamed Nomeir, Alptug Aytekin, Sennur Ulukus
TL;DR
The paper defines QXBTPIR, a quantum variant of X-secure, B-Byzantine, T-colluding PIR, where N entangled databases store K messages and up to B servers may perform arbitrary single-qudit quantum operations. It develops a CSA-based quantum coding scheme within the N-sum box framework to retain a superdense coding gain and ensure decodability under Byzantine interference, with explicit rate characterizations across regimes and extensions to eavesdroppers. A two-instance quantum CSA construction, dual QCSA matrices, and a stability analysis against interference enable reliable retrieval while identifying Byzantine behavior via consistency checks. The generalized error analysis shows any single-qudit operation can be decomposed into generalized Pauli operators, and through Kraus discretization the scheme remains robust to arbitrary Byzantine errors, highlighting strong resilience and potential practical impact for secure quantum distributed retrieval. Overall, the work advances quantum private information retrieval by quantifying Byzantine capabilities, providing concrete achievability schemes, and leveraging quantum resources to achieve doubling gains in capacity under suitable conditions.
Abstract
We consider the problems arising from the presence of Byzantine servers in a quantum private information retrieval (QPIR) setting. This is the first work to precisely define what the capabilities of Byzantine servers could be in a QPIR context. We show that quantum Byzantine servers have more capabilities than their classical counterparts due to the possibilities created by quantum encoding procedures. We focus on quantum Byzantine servers that can apply any reversible operation on their individual qudits. In this case, Byzantine servers can generate any error, i.e., this covers \emph{all} possible single qudit operations that can be applied by Byzantine servers on their qudits. We design a scheme based on cross-subspace alignment (CSA) and we show that this scheme achieves superdense coding gain in some cases.
