This paper presents an algorithm for generating synthetic LRD traffic traces, with given sample mean and Hurst parameter. The algorithm is based on the multifractal wavelet model (MWM) proposed by Riedi et al in 1999. We use a binomial conservative cascade within an iterative algorithm that ensures the required sample statistics. The synthetic trace exhibits multifractal behavior, although we do not try to capture specific multi-fractal statistics yet. As an example application, we investigate the effects of variability on the estimation of cross-traffic over point-to-point links according to the sample statistics of synthetic traces. We noticed that the correlation between cross-traffic and active probe packet dispersion measurements increases with H.