Rare-event simulation of non-Markovian queueing networks using a state-dependent change of measure determined using cross-entropy

Pieter-Tjerk de Boer

Research output: Contribution to conference › Paper › Academic

12 Downloads (Pure)

Abstract

A method is described for the efficient estimation of small overflow probabilities in non-Markovian queueing network models. The method uses importance sampling with a state-dependent change of measure, which is determined adaptively using the cross-entropy method, thus avoiding the need for a detailed mathematical analysis of the model. Experiments show that the method can be used to estimate overflow probabilities in a two-node tandem queue network model for which simulation using a state-independent change of measure does not work well.
Original languageEnglish
Pages1-6
Number of pages6
Publication statusPublished - 7 Sep 2004
Event5th International Workshop on Rare Event Simulation and Combinatorial Optimization, RESIM/COP 2004 - Budapest, Hungary
Duration: 7 Sep 2004 → 8 Sep 2004
Conference number: 5

Conference

Conference5th International Workshop on Rare Event Simulation and Combinatorial Optimization, RESIM/COP 2004
Abbreviated titleRESIM/COP
CountryHungary
CityBudapest
Period7/09/04 → 8/09/04

Fingerprint

entropy
simulation
mathematical analysis
method
sampling
experiment

Keywords

  • Cross-entropy method
  • Rare-event simulation
  • Queueing networks

Cite this

de Boer, P-T. (2004). Rare-event simulation of non-Markovian queueing networks using a state-dependent change of measure determined using cross-entropy. 1-6. Paper presented at 5th International Workshop on Rare Event Simulation and Combinatorial Optimization, RESIM/COP 2004, Budapest, Hungary.
de Boer, Pieter-Tjerk. / Rare-event simulation of non-Markovian queueing networks using a state-dependent change of measure determined using cross-entropy. Paper presented at 5th International Workshop on Rare Event Simulation and Combinatorial Optimization, RESIM/COP 2004, Budapest, Hungary.6 p.
@conference{ab53a00c95cf41bfb5fc2b68268a5e40,
title = "Rare-event simulation of non-Markovian queueing networks using a state-dependent change of measure determined using cross-entropy",
abstract = "A method is described for the efficient estimation of small overflow probabilities in non-Markovian queueing network models. The method uses importance sampling with a state-dependent change of measure, which is determined adaptively using the cross-entropy method, thus avoiding the need for a detailed mathematical analysis of the model. Experiments show that the method can be used to estimate overflow probabilities in a two-node tandem queue network model for which simulation using a state-independent change of measure does not work well.",
keywords = "Cross-entropy method, Rare-event simulation, Queueing networks",
author = "{de Boer}, Pieter-Tjerk",
year = "2004",
month = "9",
day = "7",
language = "English",
pages = "1--6",
note = "5th International Workshop on Rare Event Simulation and Combinatorial Optimization, RESIM/COP 2004, RESIM/COP ; Conference date: 07-09-2004 Through 08-09-2004",

}

de Boer, P-T 2004, 'Rare-event simulation of non-Markovian queueing networks using a state-dependent change of measure determined using cross-entropy' Paper presented at 5th International Workshop on Rare Event Simulation and Combinatorial Optimization, RESIM/COP 2004, Budapest, Hungary, 7/09/04 - 8/09/04, pp. 1-6.

Rare-event simulation of non-Markovian queueing networks using a state-dependent change of measure determined using cross-entropy. / de Boer, Pieter-Tjerk.

2004. 1-6 Paper presented at 5th International Workshop on Rare Event Simulation and Combinatorial Optimization, RESIM/COP 2004, Budapest, Hungary.

Research output: Contribution to conference › Paper › Academic

TY - CONF

T1 - Rare-event simulation of non-Markovian queueing networks using a state-dependent change of measure determined using cross-entropy

AU - de Boer, Pieter-Tjerk

PY - 2004/9/7

Y1 - 2004/9/7

N2 - A method is described for the efficient estimation of small overflow probabilities in non-Markovian queueing network models. The method uses importance sampling with a state-dependent change of measure, which is determined adaptively using the cross-entropy method, thus avoiding the need for a detailed mathematical analysis of the model. Experiments show that the method can be used to estimate overflow probabilities in a two-node tandem queue network model for which simulation using a state-independent change of measure does not work well.

AB - A method is described for the efficient estimation of small overflow probabilities in non-Markovian queueing network models. The method uses importance sampling with a state-dependent change of measure, which is determined adaptively using the cross-entropy method, thus avoiding the need for a detailed mathematical analysis of the model. Experiments show that the method can be used to estimate overflow probabilities in a two-node tandem queue network model for which simulation using a state-independent change of measure does not work well.

KW - Cross-entropy method

KW - Rare-event simulation

KW - Queueing networks

M3 - Paper

SP - 1

EP - 6

ER -

de Boer P-T. Rare-event simulation of non-Markovian queueing networks using a state-dependent change of measure determined using cross-entropy. 2004. Paper presented at 5th International Workshop on Rare Event Simulation and Combinatorial Optimization, RESIM/COP 2004, Budapest, Hungary.

Access to Document