Secure And Efficient Data Privacy Preservation Method Based On Multi-Cloud Synergy Architecture

The development of cloud computing helps to improve the efficiency of large data processing and reduce the users' cost, but this development has been severely hampered by data privacy protection. Because of the high cost and low efficiency, the technology based on a shared secret key or the full homomorphic encryption technology are not applicable to privacy protection in the large data processing in cloud computing environments.In the context of cloud computing, one effective solution is to use the private cloud which is safe and secure to process sensitive information and use the public cloud which is not guaranteed for safety to process non-sensitive information. But this hybrid cloud solution is not supported by the mainstream framework of parallel distributed big data processing, such as Map Reduce. To solve this problem, this paper studied how to use Map Reduce under the multi-cloud synergy architecture for privacy preservation in big data processing.Aiming at the multi-cloud synergy architecture, this paper proposes a general model based on data partition, it is one kind of privacy preservation model. In order to achieve privacy preservation, this mode uses the data partitioning technology to split the sensitive information which contains sensitive attribute from the original data, and then uses a private cloud to deal with the sensitive data and uses a public cloud to deal with the non-sensitive data. However, in the presence of known statistical probability of attribute's value, this paper proposes a model based on cryptography which is a more efficient privacy preservation model, this model uses the cryptography means to encrypt the sensitive information of the original data to achieve privacy preservation purposes.For the above two models, this paper designs the corresponding solutions respectively. Solution based on data partition proposes the approach "partition based on value/key" to make the two datasets after partition can be delivered separately to the multi cloud platform. This approach uses Bloom Filter to solve the difficulties of key set collecting and searching. Solution based on cryptography proposes the approach "multi-table substitution based on the probability" to achieve encryption for sensitive information, and according to the known statistical probability of attribute's value, it breaks the statistical regularity of sensitive attribute values and then protects the leakage of the privacy information, in this way, it makes the data can be delivered safely to the public cloud.Through theoretical analysis and experimental verification, the proposed solution based on data partition and solution based on cryptography under the multi-cloud synergy architecture effectively protect the data privacy and have a good performance in saving the resources of the private cloud.