Online Estimation for Packet Loss Probability of MMPP/D/1 Queuing by Importance Sampling

Abstract

In this paper, we propose a new method to estimate the packet loss probability of the MMPP/D/1 queuing system by Importance Sampling (IS). In order to estimate rare event we do not increase the arrival rate of traffic, but we decrease service rate of queuing packet. In [5], the authors also proposed an online estimation for the tail probability of FIFO queue length. However, the authors used arrival process is a Poisson process, it is simpler than MMPP arrival process in our method. Finally, we implement our algorithm and compare accuracy and simulation time of our experiments to the Monte Carlo method (MC) and conventional IS method.