In this paper, we propose two search schemes for a hierarchical group under an untrusted server ; A security-centered search scheme(SSIS) and an optimized efficient search scheme(ESIS) for commercial business use. We define `correlation resistance' as privacy requirement over encrypted search system and prove that SSIS can meet the notion. Also, we experimented two our proposed schemes. In the first try, the performance of both schemes was not good to use for practical business use. It was not until examining the reason of this that we learned the efficient DB schema must be applied into the search system for good performance. However, it was hard to apply efficient DB schema into SSIS because of its data structure. Hence, we applied efficient DB schema into only ESIS. The experiments show that ESIS is approximately 200 times faster than SSIS, which implies that other existing schemes are also not practical because the data structure of them is similar to SSIS. ESIS achieves real practicabilty by loosening its security, but with at least extend. Therefore, in the near future, it's required to develop keyword search system over encrypted data which is secure and applicable to efficient DB schema. In addition, we learned a lesson that works about the efficiency must consider mutual interactive operation with application layer as well as computational efficiency of a proposing scheme.
Category / Keywords: group search, encrypted data, keyword,trapdoor, index string, index list, efficient DB schema, primary key, foreign key Date: received 11 Jul 2006, last revised 11 Jul 2006, withdrawn 24 Dec 2006 Contact author: kokokzi at hanmail net Available formats: (-- withdrawn --) Note: The contens and authors of that paper will be changed. Version: 20061224:115848 (All versions of this report) Discussion forum: Show discussion | Start new discussion