卷积隐藏技术下的安全双方计算模型的安全性与复杂性分析
发布时间：2018-11-05 21:09
【摘要】：安全双方计算是指两个互相不信任的参与方之间在保护各自隐私的前提下协同参与计算,计算结果使得每一个参与方都能得到正确的输出,同时每一方只能得到自己的输入和输出,而不了解其他方的输入和输出。安全双方计算实际上是网络用户之间的一种分布式协议,它能够使用户在不泄漏各自的私人输入的同时完成合作计算的任务。经典的安全双方问题一直专注于零信息泄露的模式,虽然这是理想的安全双方模型,但在实际的安全双方计算中需要花费较大且效率不高。本文在传统安全双方的问题上利用卷积隐藏技术构建了卷积安全双方计算问题并基于此问题建立了第三方参与的卷积安全双方计算模型和需要部分信息公开情况下的安全双方计算模型,并对该模型进行安全性与复杂性的评估,提供了一种可以根据参与者对安全的容忍程度来调节安全和效率的方法。主要内容如下:1.研究了几个基本的安全双方求和与乘积问题,提出了相应的协议并对协议安全性与复杂性进行了分析。然后在传统安全双方计算问题的基础上,考虑参与者数据是一组n维数列的情况,提出了卷积安全双方计算问题。2.基于卷积安全双方计算问题建立了第三方参与的安全双方计算模型,并构造了两种不同安全需求情况下的实现协议,对协议的安全性与复杂性进行了分析,讨论了牺牲安全性提高参与者之间计算效率的可行性。3.提出了一种简单的复杂性量化标准,对卷积问题的不同协议的复杂性进行了分析与量化,提供了一种可以根据参与者对安全性需求来选择合理协议的方法。
[Abstract]:The calculation of security parties means that two parties who have no confidence in each other are involved in the calculation under the premise of protecting their privacy. The result of the calculation enables each participant to get the correct output. At the same time, each party can only get their own input and output, and do not understand the other side of the input and output. Secure two-party computing is actually a distributed protocol between network users, which enables users to complete the task of cooperative computing without revealing their personal input. The classical security side problem has been focused on the zero information leakage model. Although it is an ideal security two sides model, it needs to be expensive and inefficient in the actual computation of security parties. In this paper, we use convolutional hiding technology to construct the computing problem of convolutional security parties on the traditional security sides. Based on this problem, we establish the computing model of convolutional security parties with the participation of third parties and the disclosure of some information needed. The two sides of the security calculation model, The security and complexity of the model are evaluated to provide a method to adjust the safety and efficiency according to the tolerance of the participants to security. The main contents are as follows: 1. In this paper, several basic security summation and product problems are studied, and the corresponding protocols are proposed, and the security and complexity of the protocols are analyzed. Then, on the basis of the traditional security two-party calculation problem, considering that the participant data is a set of n-dimensional series, the convolution security two-party computation problem is proposed. 2. Based on the computation problem of convolutional security parties, the security parties' computing model with the participation of third parties is established, and two kinds of implementation protocols under different security requirements are constructed, and the security and complexity of the protocols are analyzed. The feasibility of sacrificing security to improve computational efficiency between participants is discussed. This paper presents a simple complexity quantization standard, analyzes and quantifies the complexity of different protocols for convolution problems, and provides a method for selecting reasonable protocols according to the security requirements of participants.
【学位授予单位】：贵州大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP309
，
本文编号：2313474
[Abstract]:The calculation of security parties means that two parties who have no confidence in each other are involved in the calculation under the premise of protecting their privacy. The result of the calculation enables each participant to get the correct output. At the same time, each party can only get their own input and output, and do not understand the other side of the input and output. Secure two-party computing is actually a distributed protocol between network users, which enables users to complete the task of cooperative computing without revealing their personal input. The classical security side problem has been focused on the zero information leakage model. Although it is an ideal security two sides model, it needs to be expensive and inefficient in the actual computation of security parties. In this paper, we use convolutional hiding technology to construct the computing problem of convolutional security parties on the traditional security sides. Based on this problem, we establish the computing model of convolutional security parties with the participation of third parties and the disclosure of some information needed. The two sides of the security calculation model, The security and complexity of the model are evaluated to provide a method to adjust the safety and efficiency according to the tolerance of the participants to security. The main contents are as follows: 1. In this paper, several basic security summation and product problems are studied, and the corresponding protocols are proposed, and the security and complexity of the protocols are analyzed. Then, on the basis of the traditional security two-party calculation problem, considering that the participant data is a set of n-dimensional series, the convolution security two-party computation problem is proposed. 2. Based on the computation problem of convolutional security parties, the security parties' computing model with the participation of third parties is established, and two kinds of implementation protocols under different security requirements are constructed, and the security and complexity of the protocols are analyzed. The feasibility of sacrificing security to improve computational efficiency between participants is discussed. This paper presents a simple complexity quantization standard, analyzes and quantifies the complexity of different protocols for convolution problems, and provides a method for selecting reasonable protocols according to the security requirements of participants.
【学位授予单位】：贵州大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP309
，
本文编号：2313474
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2313474.html
