基于云平台的加密数据计算方法

发布时间：2018-06-21 15:03

本文选题：云安全 + 安全朴素贝叶斯分类　；参考：《南京航空航天大学》2017年硕士论文

【摘要】：随着云计算的不断普及,多样的云服务为用户的工作生活带来了各种便利。典型的云服务是数据外包,数据拥有者将其拥有的大量数据外包到云服务器中存储,减少自身所需的存储管理花费。通常为了保护自身数据的隐私,拥有者将私有数据加密后再上传到云服务器中。同时,拥有者也希望利用云强大的计算能力对存储在云中的加密数据进行分析,获取原始数据间的关联信息。然而加密会使得这些数据分析变得十分困难。在不泄露数据隐私的前提下对加密数据进行安全计算成为当前的研究热点。本文的主要工作如下:1、研究了加密数据上的安全朴素贝叶斯分类问题。通过引入两个不共谋的云服务器模型并运用加法同态加密和安全多方计算,本文提出了一个安全朴素贝叶斯分类协议。与现有的方案相比,该协议将朴素贝叶斯的所有计算任务都外包给了云服务器,并且云服务器无法从整个计算过程中获得数据拥有者训练数据集、贝叶斯分类器参数以及用户测试样本的隐私信息。分析了该协议的安全性,并通过理论分析以及模拟实验分析了计算以及通信复杂度。2、对加密数据安全比较这一加密数据安全计算领域的基本问题进行了研究。通过利用加法同态加密以及姚式乱码电路,提出了一种高效的加密数据安全比较协议。并且基于提出的安全比较协议,设计了一个高效的安全范围查询协议。本文提出的安全比较协议和安全范围查询协议与已有的方案相比具有更高的效率,并且在半诚实模型下是安全的,能够保证输入输出数据的隐私信息。分析了这两个协议的安全性,通过理论分析以及模拟实验进行了性能分析与评估。
[Abstract]:With the increasing popularity of cloud computing, a variety of cloud services for users to bring a variety of work and life convenience. The typical cloud service is data outsourcing. The data owner outsource a large amount of data to the cloud server to reduce the cost of storage management. In order to protect the privacy of their own data, the owner encrypts the private data and uploads it to the cloud server. At the same time, the owner also wants to use the powerful computing power of the cloud to analyze the encrypted data stored in the cloud to obtain the correlation information between the original data. Encryption, however, makes this data analysis very difficult. Security calculation of encrypted data without revealing data privacy has become a hot research topic. The main work of this paper is as follows: 1. We study the classification of secure naive Bayes on encrypted data. By introducing two non-collusive cloud server models and using additive homomorphic encryption and secure multi-party computation, a secure naive Bayes classification protocol is proposed in this paper. Compared with the existing schemes, the protocol outsources all computing tasks of naive Bayes to cloud servers, and cloud servers cannot obtain data owner training data sets from the entire computing process. Bayesian classifier parameters and privacy information of user test samples. The security of the protocol is analyzed and the computational and communication complexity of the protocol is analyzed by theoretical analysis and simulation experiments. The basic problem of encryption data security comparison in the field of encrypted data security computing is studied. By using additive homomorphism encryption and Yao chaotic code circuit, an efficient encryption data security comparison protocol is proposed. Based on the proposed security comparison protocol, an efficient security range query protocol is designed. The security comparison protocol and the security range query protocol proposed in this paper are more efficient than the existing schemes, and are secure in the semi-honest model, which can guarantee the privacy information of the input and output data. The security of the two protocols is analyzed and the performance is analyzed and evaluated by theoretical analysis and simulation experiments.
【学位授予单位】：南京航空航天大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP309;TP393.09

【参考文献】