Signal can be modeled as a periodic or a nonperiodic stochastic process. Therefore to replicate a signal, we should keep the original character of the signal as well as the random character in it. One of plausible methods for doing such kind of job is bootstrap. However, we should modify the boostrap to accomodate the dependency in the series and their periodicities. As the pre bootraping we need to detect the existence of periodicities in the series. Two methods are given for detecting the existence of periodicities, i.e. the Fisher classical statistic, and the Chiu statistic. At the end we give an illustration. We used simulated data for testing and replicating a signal.