Toward watermarking compressed data in columnar database architectures

Abstract

AbstractA digital watermark is a secret pattern of bits inserted in the data objects to verify the integrity of the contents or to identify ownership. It is an effective technique for protecting various databases against data copyright infringement, piracy, and data tampering in both relational and not‐only‐Structured Query Language (NoSQL) databases including columnar database. Unfortunately, most of the existing digital watermarking schemes do not take into consideration the side effects that watermarking might have on the database's important characteristics such as data compression and overall performance. Thus, we are in need of database watermarking schemes that are tailored to the target database systems' specific architecture so that it does not interfere with the database's characteristics and performance. Columnar database is a type of NoSQL database where its data are stored in a column‐wise manner. This facilitates better query processing (by leaving the unrelated attributes on the disk) as well as higher compression ratio. In this research, we propose a distortion‐free fragile watermarking scheme for the columnar database architecture without interfering its underlying data compression scheme and its overall performance. The proposed scheme is flexible and can be adapted to various distributions of data. We tested our proposed scheme on both synthetic and real‐world data, and proved its effectiveness.