On the spread spectrum fingerprinting, it has been considered that the energy of embedded signal is dropped to 1/c of its original value when c copies are averaged. In this study, we analyze the effects of rounding error occurred at the embedding and averaging. The truncation of the number of decimals causes non-linear changes on the frequency components of fingerprint sequence. We estimate the
... [Show full abstract] actual attenuation factor through our simulation, and provide the best and worst case. In order to avoid the worst case, the countermeasure is proposed in this paper.