1: \begin{abstract}

2: Long-range dependence induced by heavy tails is a widely reported feature of

3: internet traffic. Long-range dependence can be defined as the regular

4: variation of the variance of the integrated process, and half the index of

5: regular variation is then referred to as the Hurst index.

6: The infinite-source Poisson process (a particular case of which is the $M/G/\infty$ queue)

7: is a simple and popular model with this property, when the tail of the

8: service time distribution is regularly varying. The Hurst index of the

9: infinite-source Poisson process is then related to the index of regular

10: variation of the service times. In this paper, we present a wavelet-based

11: estimator of the Hurst index of this process, when it is observed either

12: continuously or discretely over an increasing time interval. Our estimator is

13: shown to be consistent and robust to some form of non-stationarity. Its rate

14: of convergence is investigated.

15: \end{abstract}

16: