Flexible adjustment of the short-term correlation of LRD M/G/∞-based processes

Video represents a larger and larger portion of the traffic in Internet. This traffic is characterized by a high burstiness and a strong short- and long-range correlation, that is very important from the performance point of view. Particularly, the efficient generation of synthetic sample paths is fundamental because real traces are usually of limited length and lack the necessary diversity required to perform such analysis. In this paper, we focus on the M/G/∞ process due to its theoretical simplicity, its flexibility to exhibit both Short- and Long-Range Dependence and its advantages for simulation studies when compared to other types of processes, and we improve the adjustment of the short-term correlation of LRD M/G/∞-based processes adding autoregressive filters.