Block-based scheme for database integrity verification

Abstract

Databases play an important role today in every modern organization, verifying their integrity is needed. Watermarking can be used to protect the integrity of database. In this paper, we present a secure fragile embedding watermark technique to verify the authenticity of an outsourced numeric relational database. Our technique treats the watermark embedding as an optimization problem by securely inserting a single watermark bit in individual database partition and the optimal threshold is computed for watermark detection. The approach partitions the database in different groups of square matrix and modifies the database while preserving the field values usability constraints. The database group determinant value is used to compute the position of field to be marked. Furthermore, we evaluated our scheme on a real case study and results show its effectiveness. The proposed scheme can detect and localize the malicious modifications made to the database. The proposed technique is highly resilient to common attacks and it overcomes some limitations of previous approaches on fragile watermarking.