Efficient Importance Sampling Heuristics for the Simulation of Population Overflow in Feed-Forward Queueing Networks

V.F. Nicola; T.S. Zaburnenko

Efficient Importance Sampling Heuristics for the Simulation of Population Overflow in Feed-Forward Queueing Networks

V.F. Nicola
, T.S. Zaburnenko

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic

49 Downloads (Pure)

Abstract

In this paper we propose a state-dependent importance sampling heuristic to estimate the probability of population overï¬‚ow in feed-forward networks. This heuristic attempts to approximate the â€œoptimalï¿½ï¿½? state-dependent change of measure without the need for difficult analysis or costly optimization involved in other recently proposed adaptive importance sampling algorithms. Preliminary simulation experiments with a 4-node feed-forward network yield asymptotically efficient estimates, with relative error increasing at most linearly in the overï¬‚ow level, where state-independent importance sampling is demonstrably ineffective.

Original language	Undefined
Title of host publication	Proceedings of the Sixth Rare-Event Simulation Workshop, RESIM2006
Place of Publication	Bamberg
Publisher	Otto-Friedrich University
Pages	144-152
Number of pages	9
ISBN (Print)	not assigned
Publication status	Published - 8 Oct 2006
Event	6th International Workshop on Rare Event Simulation, RESIM 2006 - Bamberg, Germany Duration: 8 Oct 2006 → 10 Oct 2006 Conference number: 6

Publication series

Name
Publisher	Otto-Friedrich University

Workshop

Workshop	6th International Workshop on Rare Event Simulation, RESIM 2006
Abbreviated title	RESIM
Country/Territory	Germany
City	Bamberg
Period	8/10/06 → 10/10/06

Keywords

EWI-9039
METIS-248476
IR-63907

Access to Document

resim06

Cite this

@inproceedings{bc2dfc0bd3084d1ba7a22a8586ff6355,

title = "Efficient Importance Sampling Heuristics for the Simulation of Population Overflow in Feed-Forward Queueing Networks",

abstract = "In this paper we propose a state-dependent importance sampling heuristic to estimate the probability of population over{\"i}¬‚ow in feed-forward networks. This heuristic attempts to approximate the {\^a}€{\oe}optimal{\"i}¿½{\"i}¿½? state-dependent change of measure without the need for difficult analysis or costly optimization involved in other recently proposed adaptive importance sampling algorithms. Preliminary simulation experiments with a 4-node feed-forward network yield asymptotically efficient estimates, with relative error increasing at most linearly in the over{\"i}¬‚ow level, where state-independent importance sampling is demonstrably ineffective.",

keywords = "EWI-9039, METIS-248476, IR-63907",

author = "V.F. Nicola and T.S. Zaburnenko",

year = "2006",

month = oct,

day = "8",

language = "Undefined",

isbn = "not assigned",

publisher = "Otto-Friedrich University",

pages = "144--152",

booktitle = "Proceedings of the Sixth Rare-Event Simulation Workshop, RESIM2006",

note = "6th International Workshop on Rare Event Simulation, RESIM 2006 ; Conference date: 08-10-2006 Through 10-10-2006",

}

Nicola, VF & Zaburnenko, TS 2006, Efficient Importance Sampling Heuristics for the Simulation of Population Overflow in Feed-Forward Queueing Networks. in Proceedings of the Sixth Rare-Event Simulation Workshop, RESIM2006. Otto-Friedrich University, Bamberg, pp. 144-152, 6th International Workshop on Rare Event Simulation, RESIM 2006, Bamberg, Bavaria, Germany, 8/10/06.

Efficient Importance Sampling Heuristics for the Simulation of Population Overflow in Feed-Forward Queueing Networks. / Nicola, V.F.; Zaburnenko, T.S.
Proceedings of the Sixth Rare-Event Simulation Workshop, RESIM2006. Bamberg: Otto-Friedrich University, 2006. p. 144-152.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic

TY - GEN

T1 - Efficient Importance Sampling Heuristics for the Simulation of Population Overflow in Feed-Forward Queueing Networks

AU - Nicola, V.F.

AU - Zaburnenko, T.S.

N1 - Conference code: 6

PY - 2006/10/8

Y1 - 2006/10/8

N2 - In this paper we propose a state-dependent importance sampling heuristic to estimate the probability of population overï¬‚ow in feed-forward networks. This heuristic attempts to approximate the â€œoptimalï¿½ï¿½? state-dependent change of measure without the need for difficult analysis or costly optimization involved in other recently proposed adaptive importance sampling algorithms. Preliminary simulation experiments with a 4-node feed-forward network yield asymptotically efficient estimates, with relative error increasing at most linearly in the overï¬‚ow level, where state-independent importance sampling is demonstrably ineffective.

AB - In this paper we propose a state-dependent importance sampling heuristic to estimate the probability of population overï¬‚ow in feed-forward networks. This heuristic attempts to approximate the â€œoptimalï¿½ï¿½? state-dependent change of measure without the need for difficult analysis or costly optimization involved in other recently proposed adaptive importance sampling algorithms. Preliminary simulation experiments with a 4-node feed-forward network yield asymptotically efficient estimates, with relative error increasing at most linearly in the overï¬‚ow level, where state-independent importance sampling is demonstrably ineffective.

KW - EWI-9039

KW - METIS-248476

KW - IR-63907

M3 - Conference contribution

SN - not assigned

SP - 144

EP - 152

BT - Proceedings of the Sixth Rare-Event Simulation Workshop, RESIM2006

PB - Otto-Friedrich University

CY - Bamberg

T2 - 6th International Workshop on Rare Event Simulation, RESIM 2006

Y2 - 8 October 2006 through 10 October 2006

ER -