Reversible Watermarking For Relational Databases

With the rapid development of big data and cloud computing,the data is exploding.While massive data brings value to people,it also faces serious challenges.As a carrier for storing important information,such as government documents,transcation bills,and patients cases,relational database need a security measure to protect them especially.Malicious attackers try to copy,tamper or broadcast the important information without permission of the owner of relational database.This behavior not only seriously damages the rights of copyright owners,but also hurts the interest of the issuing unit.Meantimes,that may hit the safe development of digital information industry.Therefore,how to implement security protection for database in the network environment has become an urgent issue of information security.This thesis focuses on reversible watermark technology,and explores new approaches for security protection of relational database.The main contents are summarized as follows:(1)For numerical relational databases,this thesis proposes reversible watermarking technique for relational databases based on difference expansion and artificial bee colony algorithm.This method firstly divides total tuples into several subgroups using cryptographic hash function,and uses artificial bee colony algorithm to choose candidate attributes for watermark embedding.The watermark embedding is achieved by the operations applied the difference expansion method at last.Compared with the existing methods,this proposed method can effectively reduce the algorithm complexity,ensure the minimum distortion of the data,improve the watermark capacity,and improve the robustness of watermark.(2)At present,most of reversible watermarking techiques for relational databases are designed based on the modification of numerical values in the database to realize watermark embedding operation,that can't guarantee the data integrity of databases.Meanwhile,some watermarking methods have some disadvantages such as lower watermark capacity and weak robustness.In view of this,this thesis proposes a reversible watermarking method based on random forest and local binary pattern.This method firstly selects candidate attributes by random forest to reduce high dimension properties for watermark embedding,and then uses the local binary pattern to extract feature.Finally the watermark embedding is achieved by the operations applied to binary polar vectors and watermarks.Experimental results show that this method can not only ensure the integrity of database data,but also improve the robustness of watermark.In this thesis,two novel of reversible watermarking methods are proposed for relational databases,which can improve the robustness of reversible watermarking to against various attacks.Experimental results show that the proposed database watermarking methods in this thesis are feasible and have better properties than existing methods.