Journal of Information Security Reserach ›› 2023, Vol. 9 ›› Issue (11): 1086-.

Previous Articles     Next Articles

The Study of Fair Third Party in Typical Scenarios of Privacy Computing#br#

Li Chuang, Chen Xin, and Jin Fan   

  1. (China Financial Certification Authority, Beijing 100054)
  • Online:2023-11-06 Published:2023-11-30



  1. (中国金融认证中心北京100054)

  • 通讯作者: 李闯 研究员.主要研究方向为密码学、隐私计算、机器学习.
  • 作者简介:李闯 研究员.主要研究方向为密码学、隐私计算、机器学习. 陈欣 博士,高级工程师.主要研究方向为机器学习、同态加密. 金凡 硕士,高级工程师.主要研究方向为密码学、软件系统管理.

Abstract: In the industrial scenarios of privacy computing, the mutually untrusted parties can complete the joint computing by constructing secure multiparty computing protocols or applying homomorphic encryption techniques. To protect security of secret information, e. g. cryptological keys, lower the cost of communications between the parties, it is usual to adopt a third party, distributing the key pairs and secret information among parties and transmitting the intermediate ciphers. In this way, the implementation of privacy computing techniques can be accelerated. In addition, adopting such auxiliary third party can not only promote the security systematically, but also make the risks controllable and compatible to the scenarios requiring regulation and auditing process. Up to now, the basic properties of the adoptable trusted third party is not interpretated clear enough. In this paper, existed studies on privacy computing are reviewed from perspective of utility of third party as assistance, and different from the classical concept of trust third party, a concept of Fair Trusted Party (FTP) is proposed.

Key words: homomorphic encryption, trust third party, privacy computing, federated learning, secure multiparty computing

摘要: 在隐私计算的实际工业场景中通过构建密码学的多方安全算法与协议、采用同态加密以及联邦学习等技术使得互不信任的参与方能够共同完成联合计算,为了保护密钥及秘密信息安全、提升计算效率及降低通信代价等方面,通常引入辅助第三方完成相应的秘密信息管理、密文交换与转发,从而加快工业应用进程,进一步提升整体安全性,实现风险可控,并且在可监管可审计的特殊要求场景中也能发挥较大的作用.这些第三方角色与理论意义上的可信第三方存在显著差异,而目前尚没有从安全性角度对不同类型的辅助第三方进行明确的论述.从全新的角度根据隐私计算中第三方的作用梳理了当前典型隐私计算场景中的技术方法,并提取这些第三方角色的共性,明确其和传统可信第三方的区别,统一作出定义并进一步概括为公正第三方(fair trusted party FTP).

关键词: 同态加密, 可信第三方, 隐私计算, 联邦学习, 多方安全计算

CLC Number: