Research On Digital Watermarking Technology Of Relational Database

With the arrival of information age, Internet has become an important part of our life. But the information security is also facing a huge threat when we enjoy the benefits of the network. Currently, the database is widely used in political, economic, cultural and other fields. So database security is an important part of information security. Database watermarking technique is an effective method to protect database. The meaningful text or image is embedded into the database, which can be used to characterize the information of database owner or the database itself. The purpose is to protect the copyright or content of the database.In this paper, the digital watermarking technology for database is studied. The main work is given as follows:(1) A new database watermarking scheme based on Chinese Remainder Theorem (CRT) and Lagrange Interpolating Polynomial (LIP) is presented. First, the watermark is transformed into a large integer by the inverse operation of CRT, and then one-way Hash function is used to mark tuples in the database. According to Shamir (t, n) threshold scheme, n numerical attributes will be selected to be protected. The important data in this n attributes are related to the watermark by randomly generated polynomial. At the same time, the watermark will be divided into this n attributes. The watermark can be reconstructed by LIP when the watermark is extracted. Experimental results show that the proposed scheme can effectively guarantee the security of watermark information and expand the watermark capacity. At the same time, the experimental results demonstrate that the proposed scheme has good robustness performance.(2) Currently, most database watermarking is based on numerical data, which can tolerate some errors to embed watermark. The disadvantages are small embedding capacity and limited types. Watermarking technology based on Natural Language Processing is achieved by changing the expression of the statement but not changing its semantics. This paper present a database watermarking scheme based on Chinese syntactic variation. Firstly, the statements that can be used to embed watermark is searched in a database. Then one-way Hash function is used to mark tuples in the database. Label tuples as "0" if the primary key hash value is an even number. At the same time, label tuples as "1" if the primary key hash value is an odd number. The mark value and the binary watermark sequence are calculated by OR operation. If the result is "0", maintaining or changing the corresponding sentence to be active. On the other hand, if the result is "1", maintaining or changing the corresponding sentence to be passive. During the watermarking extraction, the active sentence is marked as "0" and the passive sentence is marked as "1". Also, the marker value and the primary key hash value are calculated by OR operation. Finally, the embedded watermark sequence can be obtained. Semantic analysis in Chinese is used in this paper. So that the scope of application in database watermarking technology has been more extensive. Therefore, this method has certain value for practical applications.(3) An intrusion detection system based on user behavior analysis and database watermarking is designed for Web applications. Because the behavior of users for the same database have a certain logic and relevance. So association rules data mining technology is used to mine user behavior patterns by the database audit data. Then this pattern as watermark is embedded into the original database. When the user accesses the database through Web, the system will matches the user’s behavior according to the content of the watermark. And stop responding to illegal operations. Compared with the existing intrusion detection technology based on association rules data mining, the user behavior pattern does not require another storage space alone, but need to store in the original database. It is difficult for an attacker to obtain a legal user’s behavior pattern, because of the secret of the watermark embedding algorithm and secret key. The proposed scheme can effectively resist the violent attack. The theoretical and experimental results show that this scheme is an effective attempt to combine different security strategies to ensure the security of database.