Simplified Tracing Algorithm of Fingerprinting Code Against Averaging Attack

摘要

When a binary fingerprinting codeword is embedded into digital contents using a robust spread-spectrum(SS) watermarking scheme, the averaging strategy plus addition of noise is one of cost-effective attacks. In the previous study, an optimal detector can be designed from the information theoretical point of view if the number of colluders and the attack strategy are known in advance. If the noise added to the pirated codeword is AWGN, the variance of the noise must be estimated in the optimal detector. In this paper, we simplify the optimal detector exploiting the characteristic of the discrete bias distribution of Nuida code. The proposed detector can calculate a correlation score directly from the distorted codeword without the estimation of the number of colluders as well as the variance of noise. Our experimental results reveal that the performance is very close to that of the optimal detector.