Research On Encryption Method For Character Data In The Database

Database encryption has become an important mechanism of securing data stored in database in the untrusted environments. However, how to query efficiently the encrypted data becomes a challenge.A character data encryption scheme is proposed to conduct fast query over the encrypted character data in database. While storing the encrypted data, not only encrypt each record, but also build the string index for sensitive field, and store them in an additional field in database to support string fuzzy query(LIKE query) and range query. Firstly, build a boolean matrix called string information matrix for every string in the sensitive field and three kinds of string information—character information, character adjacency information and string head&tail information are contained in this matrix; then compress the matrix into a string of bits(called string index). When querying data, we apply the principle of two-phase query. In the first phase query, we filter the portion of the records not related to the query conditions by checking the index field. In the second phase query, we decrypt the returning records of the first phase query and querying them again. This scheme can effectively support string fuzzy query and range query and reduce false positive rates. What's more, it can also support fuzzy query of Chinese characters.In addition, the confidentiality and efficiency of the proposed scheme is analyzed quantitatively. A tradeoff point between these two aspects is found out and different parameters—the length of string index m and the number of hash functions k used to build string index are determined.Finally, results of a set of experiments validate the functionality and performance of the proposed approach.