CONTENT CHARACTERISTIC BASED ROBUST DATABASE WATERMARKING

Abstract

With the booming of internet, the availability of the web based services including database services has become easy and economical to all. Digital data such as text, relational database, audio, video and software are intellectual property of owners but there outsourcing through the internet brings different threats like coping, modifying and then redistributing them. With wide spread use of relational database in many real-life applications, maintaining integrity and providing copyright protection is gaining keen interest of the researchers of the relational database. To cope with these issues the watermarking techniques playing a vital role in the last decade. This thesis reports the state-of-art watermarking techniques, highlights the shortcomings of these techniques. Then the proposed solution is given for dealing with some of the limitations of the previously presented techniques. The field of relational database watermarking is gaining interest and its becoming mature time by time. But the work done on relational database, having numerical values is more than non-numeric, so the proposed technique deals with non-numeric relational database. Considering the limitation of the previously proposed techniques, the given technique does not select any of tuple, portion (subset) from the database but uniformly distribute the watermark or consider the database as a whole while watermark embedding process. So the chances of subset selection decrease automatically. The proposed system takes into consideration a sample database and applies watermark to its tuples by selecting a set of attributes based on random bit selection and relevance/dependency of attributes of database. This report is concluded by providing results of several experiments performed in support of the effectiveness of proposed technique.