Research Article
A Random Reversible Watermarking Scheme for Relational Data
@INPROCEEDINGS{10.1007/978-3-031-25538-0_22, author={Qiang Liu and Hequ Xian and Jiancheng Zhang and Kunpeng Liu}, title={A Random Reversible Watermarking Scheme for Relational Data}, proceedings={Security and Privacy in Communication Networks. 18th EAI International Conference, SecureComm 2022, Virtual Event, October 2022, Proceedings}, proceedings_a={SECURECOMM}, year={2023}, month={2}, keywords={Relational data Reversible watermark Copyright Multiple verification}, doi={10.1007/978-3-031-25538-0_22} }- Qiang Liu
Hequ Xian
Jiancheng Zhang
Kunpeng Liu
Year: 2023
A Random Reversible Watermarking Scheme for Relational Data
SECURECOMM
Springer
DOI: 10.1007/978-3-031-25538-0_22
Abstract
In the era of Big Data, relational data is at risk of piracy and misuse when distributed, shared and used. The use of digital watermarking technology is a reliable way to protect the copyright of relational data. In order to protect the copyright of relational data and recover the original data, many reversible watermarking schemes have been proposed in recent years. But most of them cannot extract the watermark information completely under severe attacks. To address this problem, a random reversible watermarking scheme is proposed. Watermark embedding algorithm, watermark integrity checking algorithm, watermark detection algorithm and data recovery algorithm are designed. The watermark capacity is increased by embedding multiple watermarks in selected tuples, and the randomness of the watermark information distribution is increased by embedding unequal proportions of watermarks in different tuples. In extracting the watermark, the attacked bits are discarded to improve the accuracy of watermark detection. In addition, only a partition with complete watermark information is selected for watermark extraction. This not only improves the speed of watermark extraction, but also avoids the risk of key leakage from other partitions. The experimental results show that the complete watermark information can be extracted even when more than 90% of the tuples are under attack.
