Efficient coalition detection in traitor tracing

Abstract

In this paper we study the traitor tracing problem for re-broadcasting attack. In this attack, instead of building a pirate clone device (or program) based on their secret keys and sell the clone, the attackers want to stay anonymous by redistributing the decrypted content or the actual encrypting keys for the content. To defend against this type of re-broadcasting attack, the content and its actual encrypting key must come with different versions. In our setting, content is divided into multiple segments, each segment comes with multiple variations and each variation is differently encrypted. Each user/device can only play back one variation per segment through the content. A typical traitor tracing scheme for re-broadcasting attack involves two basic steps, assigning the key/variation to devices (assignment step) and detecting at least a traitor in the coalition when a series of pirated key/content are recovered (coalition detection step). We take into considerations of some realities that have been overlooked in existing schemes. As a result, we have designed a probabilistic coalition detection algorithm that is not only closer to real world scenarios but also more efficient than existing approaches. The traceability is defined in terms of the number of recovered pirate copies of the content needed to detect traitor(s) as a function of the number of traitors involved in a coalition. While existing schemes try to identify traitors one by one, our probabilistic algorithm can identify multiple traitors simultaneously and deduce the coalition size during tracing. Therefore, for the same number of total traitors in a coalition, our scheme allows the detection of all the traitors using less number of recovered copies. The superior efficiency of the our coalition detection algorithm made its adoption by AACS (Advanced Access Content System) content protection standards for next generation high-definition video optical disc. © 2008 Springer Science+Business Media, LLC.