Creating the Cryptograph Index by Query Frequencies in DAS Models

Database-as-a-Service(DAS)model is a new data management model which allow users to store their data at database service providers. Since data is stored in cryptograph form, the cryptograph index is a trade-off method to reduce the query time. So far, existing methods for creating cryptograph index ignore important statistical information of voluminous historical queries: query frequency of particular data in database, which can be used for achieving higher cryptograph query efficiency. In this paper, we first present a novel cryptograph index strategy that employs actual query frequency in bucket partition method. Such a strategy can significantly reduce the number of wrong hit tuples and hence improve query efficiency. Second, a query rewrite technique is proposed for our cryptograph index strategy to translate original queries into queries over cryptograph with the help of metadata for index. An example from real world is given to illustrate the higher performance of our query frequency based strategy than widely used equi-depth bucket partition method.