Efficient Rare Events Simulation of Gaussian Processes

A key feature of modern broadband networks is their ability to provide stringent Quality of Service (QoS) guarantees to different classes of users. A primary QoS parameter is the Cell Loss Probability (CLP), whose typical values can be very small and therefore difficult to estimate through standard Monte Carlo (MC) simulation, since long run times are required to achieve accurate results. In this paper we propose two different techniques to efficently simulate a single server queueing system, loaded with Long Range Dependent (LRD) traffic. The two techniques rely on dynamic Importance Sampling (IS) and Hybrid Monte-Carlo (HMC) respectively. In this extended abstract we briefly describe the proposed approaches together with the main theoretical results achieved in this work (Section 2). In Section 3, we conclude by showing some preliminary simulation results.