A 2-Secure Code with Efficient Tracing Algorithm

Abstract

Collusion secure fingerprinting is used to protect against illegal redistribution of digital documents. Fingerprints are embedded in documents to identify different copies. A group of colluders having access to multiple copies with different fingerprints may construct a pirate object with a fingerprint that cannot be traced. We consider c-secure codes with ɛ error that allow one of the c possible colluders to be traced and the chance of incorrect tracing to be at most ɛ. We consider a two layer construction consisting of an inner code and an outer structure and give new constructions for each. Important properties of our new inner code is that innocent users will never be accused and the code can be constructed for any number of codewords. This is particularly important as the number of codewords is the alphabet size of the outer structure. We will show that for the outer structure a c-traceability code, or a perfect hash family can be used and obtain the parameters of the combined code in terms of the parameters of the inner code and those of the outer structure. We apply these constructions to our new inner code and give parameters of the resulting c-secure codes.