170 likes | 273 Views
The Holey Grail A special score function for non-binary traitor tracing. Boris Škorić Jan-Jaap Oosterwijk Jeroen Doumen. Piracy. Copy protection Illegal or unofficial copy Traitor tracing. Traitor Tracing. Watermark embedding: user ID Pirate copy Watermark extraction: pirate ID
E N D
The Holey GrailA special score function for non-binary traitor tracing Boris Škorić Jan-Jaap Oosterwijk Jeroen Doumen
Piracy Copy protection Illegal or unofficial copy Traitor tracing / Department of Mathematics & ComputerScience
Traitor Tracing • Watermark embedding: user ID • Pirate copy • Watermark extraction: pirate ID • Watermark removal must be difficult • Watermark distortion must be difficult • Watermark detection must be difficult • Embed throughout the content in secret locations • Segment the content • Per segment, two versions: and / Department of Mathematics & ComputerScience
Collusion Compare several watermarked copies Locate the differences Create new watermark / Department of Mathematics & ComputerScience
Collusion-Resistant Traitor Tracing • Detecting the entire user ID should be difficult • The differences should not occur in all the watermark locations • Embed a random code • Per segment: • or • Probability distribution / Department of Mathematics & ComputerScience
Collusion-Resistant Traitor Tracing (II) Compare the collusion watermark to all user watermarks Attribute a level of suspicion to each user Accuse users that are too suspicious / Department of Mathematics & ComputerScience
Level of suspicion innocent accusation threshold guilty FP FN level of suspicion 0 / Department of Mathematics & ComputerScience
Code length The larger the coalition, the more difficult the tracing. 2003, Tardos: Possible to trace at least one of colluders, with error probabilities bounded by and using a code of length / Department of Mathematics & ComputerScience
Nice properties • Level of suspicion of innocent users • Zero mean (independent of collusion strategy) • Unit variance (independent of collusion strategy) • Level of suspicion of coalition • Mean is independent of collusion strategy / Department of Mathematics & ComputerScience
Larger alphabets • Using larger alphabets allows for better schemes. • 2011, Boesten et al.: Fingerprinting capacity increases • 2013, Oosterwijk et al.: Capacity-achieving scheme • But... mean level of suspicion of coalition does depend on the collusion strategy / Department of Mathematics & ComputerScience
Traitor Tracing Scheme • Probability distribution • Bias • Suspicion function • X is the version assigned to user • Y is the version chosen by coalition / Department of Mathematics & ComputerScience
The Grails • Any probability density function • Level of suspicion • Binary: Tardos scheme is not the only one! • Non-binary: We finally found a scheme (and more!) / Department of Mathematics & ComputerScience
Holy Grail • For large coalitions, the optimal scheme is • Binary: Tardos • Non-binary: / Department of Mathematics & ComputerScience
Holey Grail • We regret to inform you that this grail has holes • Using larger alphabets, allows for better schemes • But... • The fingerprinting rate of this grail decreases with / Department of Mathematics & ComputerScience
Conclusion • Collusion-Resistant Traitor Tracing Scheme • Level of suspicion of innocent users • Zero mean (independent of collusion strategy) • Unit variance (independent of collusion strategy) • Level of suspicion of coalition • Mean is independent of collusion strategy • Possible for non-binary alphabets, but this construction performs worse than was hoped. / Department of Mathematics & ComputerScience
Editor-in-Chief: Stefan Katzenbeisser, TU-Darmstadt EURASIP Journal on Information Security CfP: Special Issue on Deadline for submissions, February 28, 2014 Collusion Secure Fingerprinting and Transactional Watermarking Guest-editors: Boris Škorić, Eindhoven Univ. of Tech. Jan-Jaap Oosterwijk, Eindhoven Univ. of Tech. Thijs Laarhoven, Eindhoven Univ. of Tech. Jeroen Doumen, Irdeto BV Waldemar Berchtold, Fraunhofer Institute SIT Martin Steinebach, Fraunhofer Institute SIT Marcel Schäfer, Fraunhofer Institute SIT • Topics are among others • Collusion Secure Fingerprinting codes • Transactional Watermarking and Traitor Tracing • Collusion Secure Watermarking • Transactional Watermarking and Fingerprinting for data sets / sets of files • Real World reports or experiences or insights regarding the legal perspective http://jis.eurasipjournals.com/