You are looking at a specific version 20200308:223625 of this paper. See the latest version.

Paper 2018/1225

XMSS and Embedded Systems - XMSS Hardware Accelerators for RISC-V

Wen Wang and Bernhard Jungk and Julian Wälde and Shuwen Deng and Naina Gupta and Jakub Szefer and Ruben Niederhagen

Abstract

We describe a software-hardware co-design for the hash-based post-quantum signature scheme XMSS on a RISC-V embedded processor. We provide software optimizations for the XMSS reference implementation for SHA-256 parameter sets and several hardware accelerators that allow to balance area usage and performance based on individual needs. By integrating our hardware accelerators into the RISC-V processor, the version with the best time-area product generates a key pair (that can be used to generate 2^10 signatures) in 3.44s achieving an over 54x speedup in wall-clock time compared to the pure software version. For such a key pair, signature generation takes less than 10 ms and verification takes less than 6 ms, bringing speedups of over 42x and 17x respectively. This shows that embedded systems equipped with scheme-specific hardware accelerators are able to practically use XMSS. We tested and measured the cycle count of our implementation on an Intel Cyclone V SoC FPGA. The integration of our XMSS accelerators into an embedded RISC-V processor shows that it is possible to use hash-based post-quantum signatures for a large variety of embedded applications.

Metadata
Available format(s)
PDF
Category
Implementation
Publication info
Published elsewhere. Minor revision. Selected Areas in Cryptography - SAC 2019
DOI
10.1007/978-3-030-38471-5_21
Keywords
XMSShash-based signaturespost-quantum cryptographyhardware acceleratorFPGARISC-V
Contact author(s)
wen wang ww349 @ yale edu,ruben @ polycephaly org,jakub szefer @ yale edu
History
2020-03-08: last of 3 revisions
2018-12-30: received
See all versions
Short URL
https://ia.cr/2018/1225
License
Creative Commons Attribution
CC BY
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.