Hybrid approach for privacy preserving multi keyword ranked search on encrypted cloud data

Abstract

Cloud provides numerous advantages which attract a large number of users to utilize cloud storage to store their data. This leads to reduced infrastructure requirements thereby reducing the financial burden borne by the end users. Hence, there is a shift from hefty capital expenditure (CapEx) to lower operational expenditure (OpEx) in the form of monthly subscriptions. newlineVarious security challenges have emerged the moment user s data leaves the trusted premises as the control over their data is lost. To offer enhanced privacy, encryption is considered as a preferred solution. By encrypting the data, the end users are assured that the privacy of the data is maintained even if the data is compromised. newlineBut, encryption reduces the possibility of retrieving the relevant data without decrypting the entire collection. To provide search facility, various Searchable Encryption (SE) schemes emerged. But, these schemes either search the documents with all the search terms quickly or they can search the documents containing the subset of the search terms at high search time. Thus, it is desirable to propose a system which can overcome this imbalance between search efficiency and search capability. newlineIn this thesis, in the first phase, two conjunctive search schemes have been proposed which focus on decreasing the candidate documents to be explored. As search time is directly proportional to the comparisons performed, so with a decrease in the candidate documents, the search time decreased. The first scheme is based on document clustering and is restricted by the keywords supported per cluster. The second scheme is based on keyword binning and is independent of such restriction. Thus, efficient retrieval with limited capability is explored. newlineIn the second phase, a disjunctive search scheme has been proposed which can find the documents with a subset of the query terms. Further, with keyword binning incorporated, the relevant documents can be efficiently retrieved.