Title

Threshold-based preemption scheme for improving throughput in OBS networks

Authors

Authors

B. Zhou;M. A. Bassiouni

Comments

Authors: contact us about adding a copy of your work at STARS@ucf.edu

Abbreviated Journal Title

Photonic Netw. Commun.

Keywords

OBS networks; Performance evaluation; Burst preemption; Network; throughput; BURST-SWITCHED NETWORKS; QOS SUPPORT; SEGMENTATION; DIFFERENTIATION; SERVICE; Computer Science, Information Systems; Optics; Telecommunications

Abstract

A threshold-based scheme to improve the throughput of optical burst switching networks based on burst preemption is presented. A larger burst can preempt the channel resource reserved by a smaller burst if the size difference is equal to or larger than a threshold. Extensive simulation tests to analyze the performance of this scheme have been conducted using ring, mesh-torus, and the US LongHaul topologies. The tests used the shortest path routing algorithm with just-in-time signaling and operated at different load conditions and threshold values. The tests revealed that the "unrestricted multiple-preemption per wavelength" variant of the scheme gives nearly the same throughput performance as the "restricted single-preemption per wavelength" variant. The improvement of the network throughput depends on the value of the threshold and the value of the threshold that gives maximum throughput increases with the network load. Selecting the candidate burst with the smallest size for preemption gives better throughput than selecting a random burst. An analytical model is developed to compute the throughput of ring networks for the "restricted single-preemption per wavelength" scheme with random burst selection. The ring analytical model is quite accurate and gives results very close to those obtained by simulation.

Journal Title

Photonic Network Communications

Volume

24

Issue/Number

1

Publication Date

1-1-2012

Document Type

Article

Language

English

First Page

12

Last Page

21

WOS Identifier

WOS:000305533300002

ISSN

1387-974X

Share

COinS