Anti-collusion forensics of multimedia fingerprinting using orthogonal modulation

Area of Science:

• Digital Rights Management
• Information Security
• Multimedia Forensics

Background:

• Digital fingerprinting embeds unique identifiers in multimedia to trace unauthorized use.
• Collusion attacks combine multiple content copies to remove or weaken fingerprints, hindering tracing.

Purpose of the Study:

• To investigate the collusion resistance of a digital fingerprinting system using Gaussian distributed fingerprints and orthogonal modulation.
• To introduce and analyze novel detectors for identifying colluders.

Main Methods:

• Analysis of collusion resistance against averaging collusion attacks.
• Derivation of bounds for the maximum number of colluders (K(max)).
• Development and experimental validation of maximum and thresholding detectors for colluder identification.

Main Results:

• Fingerprinting systems are vulnerable to collusion attacks involving dozens of independent copies.
• Proposed detectors demonstrate robustness against various collusion attack types.
• A likelihood-based approach for estimating the number of colluders was developed.

Conclusions:

• Digital fingerprinting systems face significant challenges from sophisticated collusion attacks.
• The developed detectors offer improved capabilities for identifying unauthorized content use in the presence of collusion.
• Experimental results confirm the practical performance of the proposed detection methods.