Paper 2020/1373

Transciphering, using FiLIP and TFHE for an efficient delegation of computation

Clément Hoffmann, Pierrick Méaux, and Thomas Ricosset

Abstract

Improved filter permutators are designed to build stream ciphers that can be efficiently evaluated homomorphically. So far the transciphering with such ciphers has been implemented with homomorphic schemes from the second generation. In theory the third generation is more adapted for the particular design of these ciphers. In this article we study how suitable it is in practice. We implement the transciphering of different instances of the stream cipher family FiLIP with homomorphic encryption schemes of the third generation using the TFHE library. We focus on two kinds of filter for FiLIP. First we consider the direct sum of monomials, already evaluated using HElib and we show the improvements on these results. Then we focus on the XOR-threshold filter, we develop strategies to efficiently evaluate any symmetric Boolean function in an homomorphic way, allowing us to give the first timings for such filters. We investigate different approaches for the homomorphic evaluation: using the leveled homomorphic scheme TGSW, an hybrid approach combining TGSW and TLWE schemes, and the gate boostrapping approach. We discuss the costs in time and memory and the impact on delegation of computation of these different approaches, and we perform a comparison with others transciphering schemes.

Metadata
Available format(s)
PDF
Category
Implementation
Publication info
Published elsewhere. Indocrypt 2020
Keywords
Homomorphic EncryptionTFHEImproved Filter PermutatorTransciphering
Contact author(s)
clement hoffmann @ uclouvain be
pierrick meaux @ uclouvain be
thomas ricosset @ thalesgroup com
History
2020-11-02: received
Short URL
https://ia.cr/2020/1373
License
Creative Commons Attribution
CC BY

BibTeX

@misc{cryptoeprint:2020/1373,
      author = {Clément Hoffmann and Pierrick Méaux and Thomas Ricosset},
      title = {Transciphering, using {FiLIP} and {TFHE} for an efficient delegation of computation},
      howpublished = {Cryptology {ePrint} Archive, Paper 2020/1373},
      year = {2020},
      url = {https://eprint.iacr.org/2020/1373}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.