Database Watermarking Research Based On Genetic Algorithm

With the rapid development of web-based information technology,the problem of database security has become more and more urgent,for example,to deter data theft,to prevent illegal copying and to authenticate copyright,etc.In order to protect the copyright of databases,digital watermarking technology was introduced.In this paper,mechanisms to protect the copyright of a database using Genetic Algorithm are presented.Based on analysis of current database watermarking algorithms,the watermarking of relational databases is formulated as a constrained optimization problem,and Genetic Algorithm is introduced to get the optimizing solution.The main contributions of this thesis are listed as follows:First,an overview of basic theories of watermarking relational databases is given,the feasibility of introducing Genetic Algorithm into watermarking databases is studied,and basic idea of Genetic Algorithm based database watermarking algorithm is presented.Secondly,a new relational database watermarking algorithm based on is presented.The copyright owners' information is chosen as the original watermark,Standard Genetic Algorithm is applied to optimize signal processing to generate a watermark signal.The watermark then is inserted into a database and it can be extracted by watermark detection algorithm when copyright conflicts occurred.Finally,hybrid Genetic Algorithm based database watermarking algorithm is proposed. Since the local search ability of standard Genetic Algorithm is rather weak,the hybrid genetic algorithm is introduced to enhance local search ability.And the corresponding watermark insertion algorithm and watermark detection algorithm are described in detail.This article focuses on two database watermarking algorithms,namely,standard Genetic Algorithm based watermarking algorithm and hybrid Genetic Algorithm based watermarking approach.Experiments showed that the proposed database watermark embedded in a flexible program with a convenient.These techniques are resilient to tuples deletion,insertion and other attacks.