Welcome to the resource topic for 2025/1952
Title:
KPIR-C: Keyword PIR with Arbitrary Server-side Computation
Authors: Ali Arastehfard, Weiran Liu, Qixian Zhou, Zinan Shen, Liqiang Peng, Lin Qu, Shuya Feng, Yuan Hong
Abstract:Private Information Retrieval (PIR) enables clients to retrieve data from a server without revealing their query. Keyword PIR (KPIR), an extension for keyword-based queries that enables PIR using keywords, is crucial for privacy-preserving two-party analytics in unbalanced settings. However, existing KPIR solutions face two challenges in efficiently supporting arbitrary server-side computations and handling mismatched queries non-interactively.
To our best knowledge, we take the first step to introduce Keyword PIR with Computation (KPIR-C''), a novel PIR primitive that enables arbitrary non-interactive computation on responses while preserving query privacy. We overcome the arbitrary computation challenge by introducing TFHE into KPIR, which ensures efficient bootstrapping and allows arbitrary server-side computations. We address the mismatch challenge by identifying an important KPIR-C subroutine, referred to as KPIR with Default (KPIR-D’'), to remove disturbance of the computation caused by the mismatched responses. We instantiate KPIR-C with two constructions, one based on constant-weight codes and the other on recent LWE-based KPIR approaches. Both constructions enable efficient post-computation and offer trade-offs between communication overhead and runtime. Experiments show that our implemented constructions achieve competitive performance, and in some cases even outperform state-of-the-art KPIR solutions that do not support arbitrary computation.
ePrint: https://eprint.iacr.org/2025/1952
See all topics related to this paper.
Feel free to post resources that are related to this paper below.
Example resources include: implementations, explanation materials, talks, slides, links to previous discussions on other websites.
For more information, see the rules for Resource Topics .