Towards Practical Schemes For Searching The Encrypted Cloud Data

Due to the great advances in cloud computing and Internet technologies, data ownershave been motivated to outsource the storage of their data to remote cloud servers to enjoygreat data management service with an efficient cost. However, cloud computing hasintroduced new security risks that need to be addressed well to accelerate the wide adoptionfor this new technology. For security purposes, data owners usually have to encrypt theirdata prior to outsourcing it to the untrusted cloud servers. But traditional encryptionschemes are inadequate since they make indexing and searching the encrypted data morechallenging tasks. Recently, several schemes have been provided to enable searching overthe encrypted data. Such schemes are called searchable encryption (SE) schemes.However, these schemes have several limitations that need to be solved to make theapplication of such schemes more practical to the users of the cloud computing. First, themajority of the current SE schemes are limited to handling an exact search, not a similaritysearch; but the latter is an important need for all the current information retrieval systems.Second, the current similarity SE schemes are impractical for real world applicationsbecause these systems require touching the entire index during the search process, as well asincurring substantial storage cost. Third, the majority of SE schemes are limited to handlingeither a single keyword search or a Boolean search but not a multi-keyword ranked search, amore efficient model to retrieve the top documents corresponding to the provided keywords.Fourth, these systems are designed to search a single data source.In this dissertation, four methods to mitigate the above listed limitations have beenpresented. The first method allows the cloud server to answer even the mistyped queries,while preserving the privacy of the query trapdoor. The work presented in this dissertationfocuses on constructing a flexible secure index that allows the cloud server to perform theapproximate search operations without revealing the content of the query trapdoor or theindex content.The second method proposes an efficient yet secure scheme to search the encryptedcloud data, while recovering the misspellings and typographical errors that exist frequentlyboth in the search request and in the source data. The metric spaces and embedding methods have been used to construct a tree-based index, which allows retrieving only the relevantentries with a minimum number of distance evaluations.The third method proposes a secure multi-keyword ranked search scheme over theencrypted cloud data. Such scheme allows an authorized user to retrieve the most relevantdocuments in a descending order, while preserving the privacy of his search request and thecontents of documents he retrieved.The fourth method allows the authorized user to search the distributed servers withmulti-keyword queries and retrieve the most relevant documents in a descending order withrespect to their relevance to the query, while preserving the privacy of his search request andthe contents of documents he retrieved.Several experiments have been conducted to illustrate the performance of ourproposed methods. Furthermore deep analysis have been presented to prove the security ofour methods.