Debug and Analysis of Fully Homomorphic Encryption Library Based on GPU

doi:10.3969/j.issn.1671-1122.2019.06.010

Abstract

Abstract:

Fully homomorphic encryption can solve the privacy protection problem in cloud computing well, but the low efficiency is still the bottleneck of the practical application of full homomorphic encryption. There are a large number of independent matrix and vector operations in the lattice-based homomorphic encryption scheme and GPU is suitable for processing large number of independent data operations, which can greatly improve the homomorphic operation efficiency of the homomorphic encryption scheme. This paper analyzes the structure of the homomorphic encryption algorithm, verifies the reliability of the homomorphic encryption software library and analysis of noise changes and correctness of different parameters during the bootstrap process. The homomorphic encryption software library TFHE and its corresponding GPU version cuFHE was debugged and analyzed separately. The experimental results show that the GPU version of the TFHE operation speed is 4.5 times that of the CPU version TFHE. The GPU can greatly improve the homomorphic running speed of the homomorphic encryption scheme.

Key words: fully homomorphic encryption, TFHEW, GPU, code library debugging

CLC Number:

TP309

Wenchao LIU, Feng PAN, Xiaoyuan YANG, Tanping ZHOU. Debug and Analysis of Fully Homomorphic Encryption Library Based on GPU[J]. Netinfo Security, 2019, 19(6): 76-83.

Figures/Tables 9

References 13

[1]	WANG Yongjian, ZHANG Jian, CHENG Shaoyu, et al.An Improved Design of Homomorphic Encryption for Cloud Computin[J]. Netinfo Security, 2017, 17(3): 21-26.
	王永建,张健,程少豫,等.面向云计算的同态加密改进设计[J].信息网络安全,2017,17(3):21-26.
[2]	HUANG H, GONG T, CHEN P, et al.Secure Two-party Distance Computation Protocol Based on Privacy Homomorphism and Scalar Product in Wireless Sensor Networks[J]. Tsinghua Science and Technology, 2016, 21(4): 385-396.
[3]	LV Haifeng, DING Yong, DAI Yanhong, et al.Survey on LWE-based Fully Homomorphic Encryption Scheme[J]. Netinfo Security, 2015, 15(1): 32-38.
	吕海峰,丁勇,代洪艳,等. LWE上的全同态加密方案研究[J].信息网络安全,2015,15(1):32-38.
[4]	GENTRY C.Fully Homomorphic Encryption Using Ideal Lattices[C]//ACM. The 41st ACM Symposium on Theory of Computing, May 31-June 2 , 2009 Washington, DC, USA. New York: ACM, 2009: 169-178.
[5]	ZHOU Tanping, YANG Xiaoyuan, WEI Zhang, et al. Efficient Fully Homomorphic Encryption with Circularly Secure Key Switching Process[EB/OL]. , 2018-3-1.
[6]	GENTRY C, SAHAI A, WATERS B.Homomorphic Encryption from Learning with Errors: Conceptually-simpler, Asymptotically-faster, Attribute-based, Advances in Cryptology-CRYPTO 2013[M]. Berlin Heidelberg: Springer, 2013.
[7]	CHILLOTTI I, GAMA N, GEORGIEVA M, et al.Faster Fully Homomorphic Encryption: Bootstrapping in Less than 0.1 seconds[C]//Springer. Advances in Cryptology-ASIACRYPT 2016: 22nd International Conference on the Theory and Application of Cryptology and Information Security, December 4-8, 2016, Hanoi, Vietnam. Berlin Heidelberg : Springer, 2016: 3-33.
[8]	ZHOU T P, YANG X Y, LIU L F, et al. Faster Bootstrapping with Multiple Addends[EB/OL]. , 2017-3-15.
[9]	WANG W, HU Y, CHEN L, et al.Accelerating Fully Homomorphic Encryption Using GPU[C]//IEEE. High PERFORMANCE Extreme Computing, September 10-12, 2013, Waltham, USA. New York: IEEE, 2013: 1-5.
[10]	Dai W, Doröz Y, Sunar B.Accelerating NTRU based homomorphic encryption using GPUs[C]//IEEE. IEEE High PERFORMANCE Extreme Computing Conference, September 15-17, 2015, Waltham, MA USA. Waltham, MA USA. New York: IEEE, 2015: 1-6.
[11]	DAI W, SUNAR B. cuHE: A Homomorphic Encryption Accelerator Library[C]//Springer. International Conference on Cryptography and Information Security in the Balkans, September 3-4, 2010, Koper, Slovenia. Berlin Heidelberg : Springer, 2015: 169-186.
[12]	TFHE. Fast Fully Homomorphic Encryption Library over the Torus[EB/OL]. , 2017-5-2.
[13]	DAI W. CUDA-accelerated Fully Homomorphic Encryption Library[EB/OL]. , 2018-3-1.

内存	访问权限	访问范围	生命周期
寄存器内存	R/W	单线程	线程
常量内存	R	所有线程?主机	应用
纹理内存	R	所有线程?主机	应用
共享内存	R/W	块中的所有线程	块
本地内存	R/W	单线程	线程
全局内存	R/W	所有线程?主机	应用

算法	KeyGen	Enc	Dec	Nand
TFHE	2837.00 ms	0.023 ms	1.07 ms	13.00 ms
cuFHE	2758.32 ms	0.034 ms	0.045 ms	2.90 ms