Fingerprinting technique enables us to identify illegal users from a pirated copy of multimedia content. In this study, I developed an efficient detector based on signal processing techniques involving a spread spectrum method. Even if more than 100 illegal users collude, the proposed detector can detect almost all of them with reasonable computational costs. The probability that innocent users are accused by chance is limited to be extremely small. I also developed an optimal detection algorithm for a collusion secure code under a realistic threat of collusion attack. Since the optimal algorithm requires the number of illegal users and their attack straregy in advance, it is difficult to realize in essense. In order to make it practical, I introduced a statistical analysis and approximations at the detection algorithm. As the results, I succeedded to realize a simplified detection algorithm which performance is remarkably close to the optimal one.