A Lower Bound on the Joint Capacity of Digital Fingerprinting Codes Using Score Functions Based on Log-Likelihood Ratio

Digital fingerprinting codes are embedded in licensed digital contents for preventing illegal distribution. Digital fingerprinting codes are usually designed so that with high probability we can identify a part (or all) of the guilty users who collude to generate a pirated copy. In this paper we first consider the situation where the number of guilty users is equal to two and their attack model is known. We give a lower bound on the joint capacity of the digital fingerprinting codes. The lower bound is expressed as the expectation of the mutual information and is obtained by using certain score functions based on the log-likelihood ratio [12]. This result can be extended to a more general case where the number of guilty users is equal to t >= 2.