An Analysis of Database Watermarking Security

In this paper, we study watermarking security in the context of relational databases. As opposed to robustness which refers to the difficulty of removing an embedded mark without destroying the quality of the host content, security refers to the information that leaks about the secret from watermarked content. We identify two attack models, Multiple Keys Single Database and Single Key Multiple Databases as relevant attack models in which to study security of relational databases watermarking. Then, we provide a detailed analysis of watermarking security for the AHK algorithm.