A simple algorithm for generating spectrally colored, non-Gaussian signals

This work describes a simple method for generating signals conforming to a stationary random process for which the practitioner specifies both the power spectral density function and the marginal probability density function. The general approach is to first create a Gaussian random process with the appropriate spectral density and then apply a memoryless nonlinear transformation to achieve the desired marginal density. The transformation is not specified a priori but rather is simulated via an iterative “shuffling” procedure. The method is very simple to implement and yields results that are comparable to some of the more complicated methods.