Bias equalizer for binary probabilistic fingerprinting codes

Abstract

In binary probabilistic fingerprinting codes, the number of symbols "0" and "1" is generally balanced because of the design of the codeword. After a collusion attack, the balance of symbols is not always assured in a pirated codeword. It is reported in [8] that if the number of symbols in a pirated codeword is unbalanced, then a tracing algorithm can be improved by equalizing the unbalanced symbols. In this study, such a bias equalizer is customized for probabilistic fingerprinting codes utilizing the encoding parameters. Although the proposed bias equalizer is highly dependent on collusion strategies, it can improve the performance of traceability for some typical strategies to which the conventional bias equalizer can not be applied. © 2013 Springer-Verlag.