Multimedia Security, Volume 1

6Traitor Tracing

Teddy FURON

IRISA, University of Rennes, Inria, CNRS, France

This chapter presents the problem of traitor tracing and the most efficient solution known, Tardos codes. Tracing codes form an application overlay to the transmission layer by watermarking: a codeword is generated for each user, then inserted, by watermarking, into a confidential document to be shared. First and foremost, the chapter emphasizes the modeling of collusion when several traitors combine their copies. Thanks to this model, mathematics (statistics, information theory) give us the basic limits of this tool.

6.1. Introduction

A valuable document is distributed to a group of n users, each labeled by an integer: = [n], where [n] := {1, . . . , n}. These users are not to be trusted. Some of them may “leak” this document. Nothing can stop them, apart from a deterrent weapon: traitor tracing. Each user receives a unique personalized copy of the content. We assume that it is possible to do this personalization by inserting a unique codeword into the content using a watermarking technique. If a leak occurs, decoding the watermark of the illegal copy will reveal the identity of the traitor, exposing them to severe prosecution. Traitor tracing explores the case where several traitors secretly cooperate ...