加密数据上的计算密码学技术研究综述

Cryptographic technologies enabling computation over encrypted data

近年来机密数据被盗事件在国内外时有发生,其所带来的经济损失和社会影响持续加剧。尽管已有系统采取了若干安全措施,然而经验表明,攻击者往往能够通过各种方式攻入系统,盗取用户数据。为了从根本上保护数据机密性,应确保即使攻击者能够获取存储在服务器上数据的访问权限,数据机密性依然不受到威胁,即数据以加密形式存储在服务器上;另外,为了保持数据的可用性,应允许服务器在密文上进行计算。针对支持在加密数据上计算的密码学技术,如同态加密、函数加密、可搜索加密以及CryptDB和Monomi等应用系统进行了综述性介绍,并指出有待解决的关键问题和未来的研究方向。

In recent years, theft of confidential data is prevalent all over the world, thus bringing great costs to companies/ organizations and incurring personal data privacy breach. In order to prevent data leakage, almost all systems deployed nowadays adhere to the following strategy : try to prevent attackers from breaking into servers. The reasoning behind this strategy is that,if attackers are unable to break into the servers, then they cannot access the data stored on these servers as well. Nevertheless,experience has shown that the adversaries are still capable to find a way to break into the system and steal the data. To protect data confidentiality even against attackers who have access to all the stored data, the cryptographic community has proposed an alternative strategy, i. e. , store data in the encrypted form, compute over the encrypted data and never send the server the decryption key. In this way, even if an attacker reads all the data stored there, since the data is in the encrypted form and theserver never obtains the decryption key, with secure encryption scheme, the attacker will get no information about the plaintexts,thus protecting data confidentiality. There are several ways to compute over encrypted data and they provide different tradeoffs in the space of functionality, security, and efficiency. This paper gave a survey of researches in the literature on this topic, including homomorphic encryption, functional encryption, searchable encryption, as well as several systems designed for real-world applications, such as CryptDB and Monomi. Furthermore, it discussed some future research directions on this research topic as well.