A novel blind reversible method for watermarking relational databases

Abstract

Digital watermarking technology has been introduced in the past few years, not only to ensure the ownership of digital media, but also to ensure the integrity of those digital media. Reversible watermarking (which is also called invertible watermarking, or erasable watermarking) enables one to recover the original data after the contents have been authenticated. Such reversibility is highly desired in some sensitive database applications, e.g., in military and medical data. Permanent distortion is one of the main drawbacks of the entire irreversible relational database (RDB) watermarking schemes. In this article, we introduce a novel blind reversible watermarking method that ensures ownership protection in the field of RDB watermarking, whereas previous techniques have been mainly concerned with introducing permanent errors into the actual data, our approach ensures 100% recovery of the original database relation after the watermark has been detected and authenticated. In the proposed method, we utilize a reversible data embedding technique called prediction-error expansion on integers to achieve reversibility. The watermark detection can be completed successfully even when 70% of watermarked relation tuples are deleted. The experimental result shows that the blindness and robustness of this approach can withstand several kinds of database attacks.