Large buffer asymptotics for generalized processor sharing queues with Gaussian inputs (Q854995)

The authors first obtain the large-buffer asymptotics of two Gaussian flows that share the network capacity using generalized processor sharing discipline. It is assumed that flow 1 sends on average at a rate smaller than the guaranteed rate, and for flow 2 the following cases are distinguished: i) underload and ii) overload. The logarithmic asymptotics are derived. These results are applied to the integrated Gaussian process, to fractional Brownian motion and to a Gaussian flow interacting with an on-off flow.