Journal of Frontiers of Computer Science and Technology ›› 2020, Vol. 14 ›› Issue (10): 1702-1711.DOI: 10.3778/j.issn.1673-9418.1910061

Previous Articles     Next Articles

Identity-Based Matrix (Level) Fully Homomorphic Encryption Scheme

CHEN Hong, HUANG Jie, CHEN Honglin, WANG Runting, XIAO Chenglong, GUO Pengfei, JIN Haibo   

  1. College of Software, Liaoning Technical University, Huludao, Liaoning 125105, China
  • Online:2020-10-01 Published:2020-10-12

身份基矩阵层级全同态加密方案

陈虹黄洁陈红霖王闰婷肖成龙郭鹏飞金海波   

  1. 辽宁工程技术大学 软件学院,辽宁 葫芦岛 125105

Abstract:

With the development of quantum theory, traditional encryption algorithms are facing enormous challenges. Once quantum computers are available, existing encryption systems will no longer be secure. The lattice cipher is proven by theory to be a new type of encryption scheme, which can resist quantum algorithm. It has the property of fast computing speed and small space occupation, hence, it has become a hot issue in today??s research. The identity-based encryption scheme can improve the utilization of storage space and facilitate user public key management. Consequently, this paper based on the hard problem of learning with errors (LWE), constructs an identity-based matrix encryption scheme at first, then applies the improved key switching technology to reduce the dimension of homomorphic ciphertext to normal, and uses the mode conversion technique to reduce noise, realizing matrix homo-morphic operation, and getting an identity-based matrix (level) fully homomorphic encryption scheme. The scheme uses matrices as the storage structure, which has greatly improved the processing capability compared to other single-bit schemes. The scheme not only satisfies the addition and Hadamard product of the matrix, but also supports matrix dot multiplication. The experimental results show that compared with other methods, the improved key switching technology in this paper inputs less noise. Through the difficulty of the LWE problem, it is proven that the scheme satisfies IND-sID-CPA (indistinguishability-selective identity-chosen plaintext attack).

Key words: lattice cipher, learning with errors (LWE), identity-based encryption, fully homomorphic encryption

摘要:

随着量子理论的发展,传统的加密算法面临着巨大的挑战,一旦量子计算机问世,现有的加密体制将不再安全。格密码是被理论证明能抵抗量子算法的新型加密方案,它具有运算速度快,占用空间小等特点,因此成为了当今研究的热点问题。身份基加密方案,能提高对存储空间的利用率,方便用户进行公钥管理。基于格上误差学习(LWE)困难问题,首先构造一个基于身份的矩阵加密方案,然后采用改进密钥转换技术将同态运算密文的维度缩减到正常值,并使用模转换技术减小噪声,实现矩阵同态运算,得到一个基于身份的矩阵层级全同态加密方案。方案以矩阵为存储结构,较其他单比特方案,在处理能力上有了大幅提高。方案不仅满足矩阵的加法与哈达玛积,还支持矩阵点乘运算。实验结果表明,该改进密钥转换技术较其他方法引入的噪声更小。最后,通过LWE问题的困难性,证明了该方案安全性满足IND-sID-CPA。

关键词: 格密码, 格上误差学习(LWE), 身份基加密, 全同态加密