Paper 2024/1894
A non-comparison oblivious sort and its application to private k-NN
Abstract
This paper introduces a novel adaptation of counting sort that enables sorting of encrypted data using Fully Homomorphic Encryption (FHE). Our approach represents the first known sorting algorithm for encrypted data that does not rely on comparisons. The implementation leverages some basic operations on TFHE's Look-Up-Tables (LUT). We have integrated these operations into RevoLUT, a comprehensive open-source library built upon tfhe-rs, which can be of independent interest for oblivious algorithms. We demonstrate the effectiveness of our Blind Counting Sort algorithm by developing a top-$k$ selection algorithm and applying it to privacy-preserving $k$-Nearest Neighbors classification. This proves to be approximately 5x faster than current state-of-the-art methods.
Metadata
- Available format(s)
- Category
- Applications
- Publication info
- Preprint.
- Keywords
- PrivacyHomomorphic encryptionTFHEOblivious algorithmSortk-Nearest NeighborsLook-Up-Table
- Contact author(s)
-
azogagh sofiane @ courrier uqam ca
killijian marc-olivier 2 @ uqam ca
larose-gervais felix @ courrier uqam ca - History
- 2024-11-22: approved
- 2024-11-21: received
- See all versions
- Short URL
- https://ia.cr/2024/1894
- License
-
CC BY
BibTeX
@misc{cryptoeprint:2024/1894, author = {Sofiane Azogagh and Marc-Olivier Killijian and Félix Larose-Gervais}, title = {A non-comparison oblivious sort and its application to private k-{NN}}, howpublished = {Cryptology {ePrint} Archive, Paper 2024/1894}, year = {2024}, url = {https://eprint.iacr.org/2024/1894} }