Improvement of the Search over Encrypted Data in Cloud Computing byEnhancementof the Mutable Order Preserving Encryption Model

Abstract

Cloud computing has been considered as an essential technology to support future pervasive computing. It is becoming increasingly popular as it provides resources as services to its users.One of the beneficial services that the cloud provides to its user is storage. Indeed, outsourcing data to the untrusted domain leads to some vulnerability problems. To protect the outsourced data from such problems the data should be stored encrypted in the cloud. Ideally, to maintain the security of the user’s data, queries should be performed over encrypted data. Generally, various approaches support computing over encrypted data. A common approach that enables querying over encrypted data is the Ordered Preserving Encryption (OPE) scheme. It allows the sort operations (such as range query) and the comparison operation (like MIN, MAX) to be directly executed over encrypted data. Popa’s presented the first ideal-security OPE model called mutable Order-Preserving Encryption (mOPE). The ideal-security guarantee for the OPE permits the cipher-texts to leak nothing about the plain-texts besides the order. To achieve this, the mOPE model constructsthe OPE traversal tree to involve the client’s data. Moreover, to perform operations the server needs the client’s help to search over encrypted data on the OPE tree. The dependence of the server on the client to perform the requested operations dramatically produces more requests and responses between the client and the server. Concurrently, slow the system performance. This thesis presents an enhanced OPE model improved on the mOPE model to reduce the requests and responses between the client and the serverto enhance the performance of the search over encrypted data. The proposed enhanced model eliminates the dependence of the server on the client by permitting the server to perform part of the search processes without leaking any information about the original data besides the order. Moreover, to speed up the process of the search operationsit uses an indexing mechanismand developed Range_Value to order the client’s data in the server. Based on the used indexing mechanism information and the selected Range_Value the client’s data will bearranged into two types of indices, one of them follows the Popa’s technique to preserve the order of encrypted data, and the other one applies a different technique.This adds some aspect of security. The proposed enhanced model was implemented in two case study scenarios and tested on various examples,using simulation programs. Also, the mOPE model was implemented in the same case study scenarios using similar data as the enhancedOPE model.Finally, the findings from the two models were compared to evaluate the enhanced OPE model. The results shows that the enhanced OPE model succeeds in reducing therequests and responses between the client and the serveragainst the original model. The enhanced OPE model behaves better in the small Range_Value.Moreover, the experiments on the used sample of data show the best outcomes when the Range_Value is less than halve of the expectedclient’s data that will be outsourced for storage.