摘要：

数据共享过程中存在数据泄漏、信任危机等问题，且量子计算机的出现对传统加密算法带来了较大威胁。为此，提出一种基于格上密文策略属性基加密（CP-ABE）的联盟链数据共享方案。利用联盟链的准入机制以及允许存在可信第三方的特性解决在分布式网络中数据共享双方相互不信任的问题。引入基于环上容错学习的CP-ABE技术来抵御量子攻击，同时改进访问树的生成方式，将属性分为高敏感、低敏感两类，实现对数据的分级加密以保证数据的安全性。基于演化博弈论构建数据共享模型，对共享双方在联盟链数据共享体系中的选择策略进行求解和分析，并探究不同参数对演化结果的影响。实验结果表明，当属性数量呈指数级增长时，该方案的启动算法、加密算法以及解密算法的效率比基于合数阶双线群的CP-ABE方案分别提高81.6%、43.8%和56.0%，每增加一个背书节点能够使系统在执行增加用户与查询用户函数时效率分别提高36.8%与6.4%。此外，对演化模型进行模拟，结果表明，当数据固有收益、损失概率与损失收益的乘积都提高时，参与联盟链带来的收益提高，用户更倾向于参与联盟链。

关键词: 数据共享, 密文策略属性基加密, 环上容错学习, 联盟链, 演化博弈论

Abstract:

In the process of data sharing, problems of data leakage and trust crisis can arise, and with the emergence of quantum computers, the threat to traditional encryption algorithms is greater. To this end, this paper proposes a consortium blockchain data sharing scheme that uses lattice Ciphertext Policy Attribute-Based Encryption(CP-ABE) to solve the problem of mutual distrust between two data-sharing parties in a distributed network, using the access mechanism of the consortium blockchain and the characteristics that allow the existence of a trusted third party. CP-ABE based on Ring-Learning with Errors(R-LWE) is introduced to resist quantum attacks and improve access tree generation to classify attributes into two categories of high and low sensitivity, to ensure data security through hierarchical encryption of the data. The data sharing model is constructed based on an evolutionary game, whereby the selection strategy of the sharing parties in the consortium blockchain data sharing system and the influence of different parameters on the evolution results are analyzed. The experimental results show that as the number of attributes increases exponentially, the efficiency of the hierarchical encryption scheme in the SystemSetup, DOEncrypt, and DUDecrypt algorithms become 81.6%, 43.8%, and 56.0% higher than that of the CP-ABE scheme which is based on composite order bilinear groups. Each addition of an endorsement node can increase the system efficiency of executing the functions of adding users and querying users by 36.8% and 6.4%, respectively. In addition, simulations of the evolutionary model show that, the inherent benefits of data and the product of loss probability and loss income increase, thereby increasing the income brought by participating in the consortium blockchain, making users more inclined to join the consortium blockchain.

Key words: data sharing, Ciphertext Policy Attribute-Based Encryption(CP-ABE), Ring-Learning with Errors(R-LWE), consortium blockchain, evolutionary game theory