Uniform approximations for the \(M/G/1\) queue with subexponential processing times (Q543561)

From the authors' abstract and the introduction: ``The asymptotic behaviour of the distribution of an \(M/G/1\) queue with subexponential service time distribution and FIFO (first in, first out) scheduling are investigated. The steady state waiting time distribution of this queue is described by the so-called heavy-traffic approximation and heavy-tail asymptotic. In the main result of this paper, the observation is made that the behaviour of the studied waiting-time for service time distributions decaying slower than subexponentially includes a third region where neither the heavy-traffic approximation nor the heavy-tail asymptotic are valid. The authors provide details of the used approximation of the steady-state waiting time for which they derive uniform expressions of the minimum conditions for the service time distribution. This approach allows them to draw a connection between the asymptotic behaviour of the steady-state waiting time distribution and that of the associated random walk. The obtained results can be of practical use for numerical approximations of the steady-state waiting time distribution. The results are confirmed by some simulations, the main idea of which is based on the numerical computation of the integrated tail distribution of these processing times and their first few moments. In all these experiments the heavy-tail asymptotic holds tend to the mean and variance of the integrated tail distribution.''