Fragile Watermarks for Database Relations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current digital watermarking technologies fail to effectively detect and localize malicious alterations in database relations with categorical attributes, as they either introduce distortions or are not designed for categorical data, and existing schemes are not suitable for frequent updates or cannot tolerate modifications in such data types.
Innovation Solution
A fragile watermarking scheme that securely divides tuples into groups, computes primary key hashes, and sorts them to embed watermarks without altering data values, allowing for detection and localization of modifications within specific groups, ensuring invisibility and distortion-free embedding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If digital watermarking is applied to database relations with categorical attributes, then tamper detection capability is improved, but data distortion is introduced
Solution Approach 1:
The patent divides the database relation into multiple groups based on primary key hash values. Each group is watermarked independently, allowing tamper detection at the group level while preserving the integrity of individual categorical attributes. This segmentation enables localized tamper detection without requiring distortion of the entire database.
Solution Approach 2:
The patent uses sorting order as an intermediary mechanism to embed watermarks. Instead of directly modifying categorical attribute values (which would cause distortion), the system sorts tuples within groups based on computed hash values and uses this sorting order to represent watermark bits. This intermediary approach enables watermark embedding without altering the actual data values.
2Reliability
If existing digital signature schemes are used for tamper detection, then detection capability is improved, but localization capability deteriorates
Solution Approach 1:
By dividing the database into multiple groups and embedding independent watermarks in each group, the patent enables localization of tampering to specific groups. When a watermark verification fails, the system can identify which specific group contains the tampered data, providing precise localization rather than just overall detection.
3Strength
If watermark embedding modifies data values, then watermark robustness is improved, but data authenticity deteriorates
Solution Approach 1:
The patent employs sorting order as an intermediary to embed watermarks without modifying categorical attribute values. The sorting is based on computed hash values of the data, and the sorting position represents the watermark bit. This approach maintains data authenticity because the actual data values remain unchanged, while still achieving watermark embedding through the intermediary sorting mechanism.
Data Source
AI summary
Disclosed is a fragile watermarking scheme for detecting and localizing malicious alterations made to a database relation with categorical attributes without introducing distortions to cover data. A watermark for a tuple group may be inserted by selectively switching the position of tuples in tuple pairs using a tuple hash associated with each tuple in the tuple pair; and a corresponding bit in a watermark derived from the tuple group using a embedding key, a primary key and hash functions.