Optimization of Tardos's Fingerprinting Codes in a Viewpoint of Memory Amount.

It is known that Tardos's collusion-secure probabilistic fingerprinting code (Tardos code) has length of theoretically minimal order. However, Tardos code uses certain continuous probability distribution, which causes that huge amount of extra memory is required in a practical use. An essential solution is to replace the continuous distributions with finite discrete ones, preserving the security. In this paper, we determine the optimal finite distribution for the purpose of reducing memory amount; the required extra memory is reduced to less than 1/32 of the original in some practical setting. Moreover, the code length is also reduced (to, asymptotically, about 20.6% of Tardos code), and some further practical problems such as approximation errors are also considered. [ABSTRACT FROM AUTHOR]