This paper analytically studies performance improvement achieved by packet dispersion. Contrary to the previous work on packet dispersion, we consider the tail distribution of queue length and that of packet delay as performance measures, and we model a packet network as multiple parallel queues where the arrival processes from sources are not renewal but highly bursty. To appropriately evaluate the performance improvement achieved by packet dispersion, we develop approximate formulas to estimate the tail distributions. Our approximate formulas yield more accurate estimations than the standard approximate formulas. In the numerical results, we observe that packet dispersion can greatly improve the delay performance of packets. We also see that packet-level load balancing is superior to flow-level load balancing for any distribution ratio.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Fumio ISHIZAKI, "On the Performance Improvement Achieved by Packet Dispersion" in IEICE TRANSACTIONS on Communications,
vol. E89-B, no. 7, pp. 1977-1986, July 2006, doi: 10.1093/ietcom/e89-b.7.1977.
Abstract: This paper analytically studies performance improvement achieved by packet dispersion. Contrary to the previous work on packet dispersion, we consider the tail distribution of queue length and that of packet delay as performance measures, and we model a packet network as multiple parallel queues where the arrival processes from sources are not renewal but highly bursty. To appropriately evaluate the performance improvement achieved by packet dispersion, we develop approximate formulas to estimate the tail distributions. Our approximate formulas yield more accurate estimations than the standard approximate formulas. In the numerical results, we observe that packet dispersion can greatly improve the delay performance of packets. We also see that packet-level load balancing is superior to flow-level load balancing for any distribution ratio.
URL: https://global.ieice.org/en_transactions/communications/10.1093/ietcom/e89-b.7.1977/_p
Copy
@ARTICLE{e89-b_7_1977,
author={Fumio ISHIZAKI, },
journal={IEICE TRANSACTIONS on Communications},
title={On the Performance Improvement Achieved by Packet Dispersion},
year={2006},
volume={E89-B},
number={7},
pages={1977-1986},
abstract={This paper analytically studies performance improvement achieved by packet dispersion. Contrary to the previous work on packet dispersion, we consider the tail distribution of queue length and that of packet delay as performance measures, and we model a packet network as multiple parallel queues where the arrival processes from sources are not renewal but highly bursty. To appropriately evaluate the performance improvement achieved by packet dispersion, we develop approximate formulas to estimate the tail distributions. Our approximate formulas yield more accurate estimations than the standard approximate formulas. In the numerical results, we observe that packet dispersion can greatly improve the delay performance of packets. We also see that packet-level load balancing is superior to flow-level load balancing for any distribution ratio.},
keywords={},
doi={10.1093/ietcom/e89-b.7.1977},
ISSN={1745-1345},
month={July},}
Copy
TY - JOUR
TI - On the Performance Improvement Achieved by Packet Dispersion
T2 - IEICE TRANSACTIONS on Communications
SP - 1977
EP - 1986
AU - Fumio ISHIZAKI
PY - 2006
DO - 10.1093/ietcom/e89-b.7.1977
JO - IEICE TRANSACTIONS on Communications
SN - 1745-1345
VL - E89-B
IS - 7
JA - IEICE TRANSACTIONS on Communications
Y1 - July 2006
AB - This paper analytically studies performance improvement achieved by packet dispersion. Contrary to the previous work on packet dispersion, we consider the tail distribution of queue length and that of packet delay as performance measures, and we model a packet network as multiple parallel queues where the arrival processes from sources are not renewal but highly bursty. To appropriately evaluate the performance improvement achieved by packet dispersion, we develop approximate formulas to estimate the tail distributions. Our approximate formulas yield more accurate estimations than the standard approximate formulas. In the numerical results, we observe that packet dispersion can greatly improve the delay performance of packets. We also see that packet-level load balancing is superior to flow-level load balancing for any distribution ratio.
ER -