基于LWE两方数相等的保密计算协议  

Secure Computation Protocol Based on LWE Two-party Numbers Equality

作　　者：解丹[1] 杨波[1] 邵志毅[1] 徐彦蛟[1] 杜军强[1] 

机构地区：[1]陕西师范大学计算机科学学院,西安710062

出　　处：《计算机工程》2013年第12期122-125,共4页Computer Engineering

基　　金：国家自然科学基金资助项目(61272436);广东省自然科学基金资助项目(10351806001000000)

摘　　要：保密地比较两方数是否相等是安全多方计算(SMC)问题中重要的研究内容,其在数据挖掘、在线推荐服务、在线预定服务、医药数据库等领域有着重要应用。针对半诚实模型下两方保密比较协议无法抵抗恶意攻击的问题,提出一种恶意模型下两方数相等的保密计算协议,采用基于格上差错学习(LWE)困难性问题的公钥加密机制和Paillier加密方案,使得存在恶意攻击者的情况下能够阻止恶意攻击行为发生,同时证明协议在恶意模型下是安全的。分析结果表明,该协议执行完成后不会泄露通信双方的私有信息,与半诚实模型下两方保密比较协议相比,能有效抵抗恶意攻击者的攻击,为SMC通信提供了较好的解决方案。The equation test is an important part in Security Multi-party Computation（SMC）. It has important application in the fields of data mining, recommendation service, online dating service, and medical database. According to the defects existing in the protocols of comparing two data based on security under the semi-honesty model, this paper proposes a secure computation protocol for two-party numbers equality test in the malicious model. The protocol uses the public-key encryption mechanism based on lattice Learning With Error（LWE） difficult problem and Paillier encryption scheme, it can prevent malicious attacks in the case of existing malicious attacker, and at the same time proves that agreement is safe under the malicious model. Analysis results prove that the protocol after the implementation is completed, and no private information in both communication parties is revealed. Compared with the protocols of comparing two data based on security under the semi-honesty model, the proposed protocol can effectively resist the attacks from the malicious adversary and provides a good solution for the communication with high needs.

关 键 词：安全多方计算 两方数相等 半诚实模型 恶意模型 差错学习困难性问题 Paillier加密方案 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]