An Efficient Method for Generating Autocorrelated Non-Gaussian Sequences

We propose a novel method for generating random sequences matching both a target non-Gaussian distribution and a target autocorrelation. In the proposed scheme, samples matching the target distribution are drawn independently and then suitably rearranged to match the target autocorrelation. Our method overcomes the difficulties of the standard method in generating sequences with unknown inverse distribution. Besides communications, our results find applicability in many different areas concerned with non-Gaussian processes. Many examples are included.