Accelerating lattice based proxy re-encryption schemes on GPUs

Gyana Sahu, Kurt Rohloff

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Proxy Re-Encryption (PRE) is an indispensable tool in many public-key cryptographic schemes that enables users to delegate decryption rights to other users via a proxy. In this work, we present a high performance implementation of PRE schemes on NVIDIA GPUs. We target two lattice based PRE schemes, BV-PRE and Ring-GSW PRE defined over polynomial rings. We design a parallel Number Theoretic Transform (NTT) procedure capable of working on arbitrary precision moduli (in CRT form) and demonstrate several low level and GPU optimizations techniques to accelerate the PRE schemes. For the same or higher security settings our results show 39x to 228x factors of improvement in performance with a peak throughput of 6.3 Mbps when compared to the CPU implementation of the BV-PRE scheme in the PALISADE lattice crypto software library. Similarly, for the Ring-GSW PRE scheme we achieve a peak throughput of 49 Mbps and up to 11x improvement in performance.

Original languageEnglish (US)
Title of host publicationCryptology and Network Security - 19th International Conference, CANS 2020, Vienna, Austria, December 14–16, 2020, Proceedings
EditorsStephan Krenn, Haya Shulman, Serge Vaudenay
PublisherSpringer Science and Business Media Deutschland GmbH
Pages613-632
Number of pages20
ISBN (Print)9783030654108
DOIs
StatePublished - 2020
Event19th International Conference on Cryptology and Network Security, CANS 2020 - Vienna, Austria
Duration: Dec 14 2020Dec 16 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12579 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference19th International Conference on Cryptology and Network Security, CANS 2020
Country/TerritoryAustria
CityVienna
Period12/14/2012/16/20

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Keywords

  • CUDA
  • GPU acceleration
  • Homomorphic encryption
  • Ring-LWE

Fingerprint

Dive into the research topics of 'Accelerating lattice based proxy re-encryption schemes on GPUs'. Together they form a unique fingerprint.

Cite this