Abstract: With the development of the cloud-computing technology, the volume of data generated and processed by?various applications?is becoming increasingly large. However, the security of large-scale data sets is not well ensured currently. Although cryptography is an important tool to guarantee data security, secure problems occur when conventional encryption algorithms are applied directly to?large-scale data sets. In this paper, an algorithm for encrypting large-scale data sets is?proposed. To implement this algorithm, it proposes a grading permutation algorithm for encrypting large-scale data sets. Global diffusion?can be achieved for a data set?with size of [N(2mN<2m+1)] after the data set is encrypted for?m+2 cycles. Then, the algorithm is implemented?on the Hadoop platform based on the?MapReduce programming model. Theoretical analyses and experiments show?that this algorithm features splendid global diffusibility for large-scale data set encryption. Compared with traditional algorithms, the proposed algorithm has significant advantages in parallelism and?computation efficiency.

Key words: cloud-computing, cryptography, permutation, diffusion, Hadoop platform, MapReduce

摘要： 随着云计算技术的蓬勃发展，各种应用产生的数据量日益庞大，然而目前大规模数据集的安全并未得到充分保障。密码学是确保数据安全的有效手段，但是将传统加密方法直接应用于大数据加密存在安全隐患。提出了针对大规模数据集加密的排列算法，包括两方面工作：提出一种适用于大规模数据集的分组加密排列算法，对数据规模为[2mN<2m+1]的数据集，经过[m+2]轮加密可实现全局扩散；基于MapReduce编程模型，在Hadoop平台上实现了分组加密排列算法。理论分析与实验结果表明，该排列算法具有优异的全局扩散性。

关键词: 云计算, 密码学, 排列, 扩散, Hadoop平台, MapReduce