Secure Indexes for Keyword Search in Cloud Storage

Our investigation challenge is designing an auxiliary data structure to support keyword lookups on encrypted documents stored in the cloud. We propose an extension of “Color Indexes”, a secure indexing technique introduced by Hore et al. The design utilizes false positive search results to hide sensitive information and allows users to adjust security level in trade-off with the performance by tuning system parameters. We focus mainly on improving practicality of the indexes on real world application, especially, analyzing and conducting experiments on the effects of skewed keyword distribution and document sizes. To handle skewed datasets, we recommend using a second level hash function to prevent information leakage from color statistic and using a sampling based technique for more accurate error prediction. The efficiency of our indexes is demonstrated in experiments on English Wikipedia, Reuter-21578, and Enron Email Corpus.報告番号: ; 学位授与日: 2014-09-26 ; 学位の種別: 修士 ; 学位の種類: 修士(工学) ; 学位記番号: ; 研究科・専攻: 工学系研究科・電気系工学専