Sequential Traitor Tracing

Traceability schemes allow detection of at least one traitor when a group of colluders attempt to construct a pirate decoder and gain illegal access to digital content. Fiat and Tassa proposed dynamic traitor tracing schemes that can detect all traitors if they attempt to rebroadcast the content after it is decrypted. In their scheme the content is broken into segments and marked so that a re-broadcasted segment can be linked to a particular subgroup of users. Mark allocation for a segment is determined when the re-broadcast from the previous segment is observed. They showed that by careful design of the mark allocation scheme it is possible to detect all traitors. We consider the same scenario as Fiat and Tassa and propose a new type of traceability scheme, called sequential traitor tracing, that can efficiently detect all traitors and does not require any real-time computation. That is, the marking allocation is pre-determined and is independent of the re-broadcasted segment. This is very attractive as it allows segments to be shortened and hence the overall convergence time reduced. We analyse the scheme and give two general constructions one based on a special type of function family, and the other on error correcting codes. We obtain the convergence time of these schemes and show that the scheme based on error correcting codes has a convergence time which is the same as the best known result for dynamic schemes.