Prediction-based robust blind reversible watermarking for relational databases

As the size of database grows, the possibility of database corruption also increases. One such example is of temporal databases in which deletion never occurs except in case of vacuuming. A strong security mechanism is needed to find any database modification. In case of any tampering, tampered data should be identified and recovery of original data from the tampered one is also essential. In this work, a new watermarking scheme for database authentication and forensic analysis is developed. The proposed system uses a set of watermark bits to make a validation and recovery mechanism for database authentication. In order to measure the robustness of this approach, online available yahoo financial data is watermarked through this approach and simulation of insertion, modification and deletion attacks are performed. Normalised correlation (NC) and mean square error (MSE) are used for measuring the performance of this approach. Extensive analysis shows that the proposed method is robust against various forms of database attacks, including insertion, deletion and modification. In future, in order to identify the best possible locations for embedding the watermark, optimisation algorithms can be used. Also methods may be developed for enhancing the embedding capacity of the watermark. Copyright © 2021 Inderscience Enterprises Ltd.