一种MapReduce框架下的数据隐私保护方案

A Data Privacy Protection Scheme Based on MapReduce

为了解决MapReduce框架下因私有加密密钥被发送到不信任的云服务器,造成数据隐私泄露问题。提出了一种改进的完全同态加密方案,该方案应用逻辑Merkle哈希树存储结构,降低时空代价,解决元数据动态和认证路径问题,对输入文件分解(MAP阶段)和密文检索算法(Reduce阶段)进行优化。此外,为了提高计算成本和安全性检测的效率,采用了标量同态方法,避免了繁杂的多项式时间计算。并提出了一种优化的三值搜索尝试(TST)算法来管理元数据认证。实验结果表明,优化后的密码体制能明显降低通信和计算成本,整体密文检索时间降低了3倍。

In order to solve the problem of data privacy leakage caused by private encryption keys being sent to untrusted cloud servers under MapReduce framework.An improved fully homomorphic encryption scheme is proposed.The scheme uses the logical Merkle hash tree storage structure to reduce the space-time cost,solve the problem of metadata dynamic and authentication path,and optimize the input file decomposition(MAP stage)and ciphertext retrieval(Reduce stage).In addition,in order to improve the computational cost and the efficiency of security detection,the scalar homomorphism method is used to avoid the complicated polynomial time calculation.An optimized ternary search attempt(TST)algorithm is proposed to manage metadata authentication.The experimental results show that the optimized cryptosystem can significantly reduce the communication and computing costs,and the overall ciphertext retrieval time is reduced by three times.