Research On Efficient Ranked Search Schemes Over Encrypted Cloud Data

In the times of information explosion,more and more cloud data owners are preferring to migrate their personal data to the cloud as to save storage costs,and thus enjoy more convenience.However,to effectively protect personal privacy,much of the important data should be encrypted,which makes traditional data utilization methods based on plaintext keyword search difficult to work.Therefore,the new efficient search schemes over cloud data become highly critical.In view of the efficiency issues of ranked search during the cloud environment,this thesis mainly focuses on the following points:(1)Multi-keyword ranked search based on HCB-Tree(MRSE-HCB): This scheme uses HCBTree to improve search efficiency,and builds a search tree based on binary k-means clustering algorithm which stores data in the leaf nodes.The intermediate node stores the extracted maximum vector of the cluster formed by the subtree rooted at the node.During the searching process,this scheme will first search the document most relevant to the keyword,and then search the neighboring nodes of this leaf node where the document is stored.In this way,the scheme can make the search algorithm reach sub-linear computational complexity.(2)Search efficiency optimization based on selection of the filtering threshold(OFT):Considering that the existing schemes over encrypted data based on tree index all use the filtering threshold for improving search efficiency,but the initial filtering threshold is set to zero.At the early stage of searching,the pruning filtering effect of this threshold is not obvious.Therefore,this thesis suggests a search efficiency optimization scheme with the filtering threshold selection.This scheme selects an appropriate filter threshold when the search is initiated,thereby more effectively pruning a large number of irrelevant subtrees,and further improving the pruning efficiency during the depthfirst search process.This scheme is applicable to scenarios of ranked search over encrypted data based on tree index,which effectively enhancing search efficiency with no impact on the accuracy of results.(3)Based on Eclipse development platform,the experimental simulation system of multikeyword ranked search over encrypted data is specifically designed and implemented with the combination of Java language and Swing framework.The functional design of the system is mainly divided into six modules: full document preprocessing,document data clustering,index structure generation,plaintext vector encryption,multi-keyword search and main interface.