Accelerating lattice based proxy re-encryption schemes on GPUs
Document Type
Conference Proceeding
Publication Date
1-1-2020
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.
Identifier
85098239549 (Scopus)
ISBN
[9783030654108]
Publication Title
Lecture Notes in Computer Science Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics
External Full Text Location
https://doi.org/10.1007/978-3-030-65411-5_30
e-ISSN
16113349
ISSN
03029743
First Page
613
Last Page
632
Volume
12579 LNCS
Recommended Citation
    Sahu, Gyana and Rohloff, Kurt, "Accelerating lattice based proxy re-encryption schemes on GPUs" (2020). Faculty Publications.  5556.
    
    
    
        https://digitalcommons.njit.edu/fac_pubs/5556
    
 
				 
					