Querying encrypted character data in DAS model

• Published in: 2010 International Conference on Networking and Digital Society Vol. 2; pp. 402 - 405
• Main Authors: Nian Liu, Yajian Zhou, Xinxin Niu, Yixian Yang
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.05.2010
• Subjects: Application software; character; Costs; Cryptography; Data engineering; Database as a Service; Filtration; fuzzy search; Indexes; Information security; Laboratories; Relational databases; Telecommunication switching
• ISBN: 1424451620; 9781424451623
• DOI: 10.1109/ICNDS.2010.5479313

There has been increasing concern about Databaseas-a-Service(DAS) architectures. In the DAS model, the conventional scheme for query over encrypted data in database is to construct efficient index in database. This can ensure query efficiency. However, the fuzzy search is often used in the query of character data. It will be very tough owing to data encryption. Even the tiny changes of plaintext, the ciphtext will be quite distinct. Therefore, it is crucial to construct efficient index of encrypted character strings. To address these problems, a cryptographic scheme of character data in relational database is proposed in this paper. In this scheme, the char or varchar data type and clob data type will be considered respectively. For the data of char or varchar type, an index based double filtration is established. The index includes two parts: the first part of index is used to determine the set of existed characters in original string, and the second part stores the characteristic value of the characters' positions. The first part of index implements the first filtration, which will filter all the records include inaccurate characters. The second part of index is used to verify the positions of character strings. It's the second filtration. For the data of clob type, instead of the original character data, the positions of all the character strings will be stored. This scheme avoids needless decryption and data transfer. Consequently, better query efficiency is achieved and fuzzy search is well supported.