Active Scheduling Algorithm for Intra-ONU Dynamic Bandwidth Allocation in Ethernet Passive Optical Networks
碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 96 === Time-division multiple access scheme on a single wavelength is more attactive for upstream traffic in EPON. Each ONU doesn’t know start time of upstream bandwidth and granted window size each other, and it may cause conflicts. Considering upstream bandwidth ut...
Main Authors: | , |
---|---|
Other Authors: | |
Format: | Others |
Language: | zh-TW |
Published: |
2008
|
Online Access: | http://ndltd.ncl.edu.tw/handle/19719649442951294115 |
id |
ndltd-TW-096NCKU5392062 |
---|---|
record_format |
oai_dc |
spelling |
ndltd-TW-096NCKU53920622015-11-23T04:03:10Z http://ndltd.ncl.edu.tw/handle/19719649442951294115 Active Scheduling Algorithm for Intra-ONU Dynamic Bandwidth Allocation in Ethernet Passive Optical Networks 在乙太被動光纖網路中以主動式排程演算法達到光纖網路單元內部動態頻寬分配 Hung-Yuan Chen 陳鴻源 碩士 國立成功大學 資訊工程學系碩博士班 96 Time-division multiple access scheme on a single wavelength is more attactive for upstream traffic in EPON. Each ONU doesn’t know start time of upstream bandwidth and granted window size each other, and it may cause conflicts. Considering upstream bandwidth utilization, packet delay and Quality-of-Service(QoS), it is very important to design an efficient bandwidth allocation algorithm conforming to requirement of QoS. OLT and ONU communicate with each other by MPCP. We use inter-scheduling and intra-scheduling algorithms to manage bandwidth allocation for inter-ONU and intra-ONU, respectively. Each ONU sends REPORT message to OLT in their transmission window, and OLT executes inter-scheduling algorithm to calculate transmission window for receiving REPORT message from all ONUs.OLT sends a GATE message which contains start time and size of transmission window to each ONU. ONU used to decide the transmission order of queue data and the transmission size of queue data by executing intra-scheduling, we call it passive intra-scheduling. The thesis focus on design of intra-scheduling algorithm, and we propose an active intra-scheduling algorithm. The general idea of active intra-scheduling algorithm is recording several queue sets in the reservation fields in REPORT message. We set thresholds for different classes to satisfy requirements of QoS. Therefore, OLT gets several sets of bandwidth requirement for receiving a REPORT message from one ONU. According to the bandwidth computed from inter-scheduling algorithm, OLT determines a appropriate allocation from queue sets in REPORT message. ONU gets granted window size based on REPORT which are send by ONU in previous cycle, so we call it active intra-scheduling algorithm. Using active intra-scheduling also eliminate unused slotted remainder (USR), besides satisfying requirement of QoS and reasonable packet delay for different class. ONU can’t completely make use of upstream bandwidth which are given by OLT because ethernet packet is not divided, and it will cause USR which is from 0 to 1517bytes. ONUs contrast granted window size in GATE message with REPORT which was sent to OLT, and it will find a queue set in which sum of queue report equal to granted window size. ONUs transmit packet according to the queue set which is found by contrasting in priority order. For this reason, ONU can completely eliminate USR. We use the inter-scheduling which is mentioned in other reference and combine it with active intra-scheduling for enhancing throughput, satifying requirement of QoS, and reasonable packet delay. Finally, we compare the proposed scheme with EAF scheduling in our simulation results. We simulate USR, throughput, packet delay for uniform and non-uniform load to prove our scheme better. Chuan-ching Sue 蘇銓清 2008 學位論文 ; thesis 73 zh-TW |
collection |
NDLTD |
language |
zh-TW |
format |
Others
|
sources |
NDLTD |
description |
碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 96 === Time-division multiple access scheme on a single wavelength is more attactive for upstream traffic in EPON. Each ONU doesn’t know start time of upstream bandwidth and granted window size each other, and it may cause conflicts. Considering upstream bandwidth utilization, packet delay and Quality-of-Service(QoS), it is very important to design an efficient bandwidth allocation algorithm conforming to requirement of QoS.
OLT and ONU communicate with each other by MPCP. We use inter-scheduling and intra-scheduling algorithms to manage bandwidth allocation for inter-ONU and intra-ONU, respectively. Each ONU sends REPORT message to OLT in their transmission window, and OLT executes inter-scheduling algorithm to calculate transmission window for receiving REPORT message from all ONUs.OLT sends a GATE message which contains start time and size of transmission window to each ONU. ONU used to decide the transmission order of queue data and the transmission size of queue data by executing intra-scheduling, we call it passive intra-scheduling.
The thesis focus on design of intra-scheduling algorithm, and we propose an active intra-scheduling algorithm. The general idea of active intra-scheduling algorithm is recording several queue sets in the reservation fields in REPORT message. We set thresholds for different classes to satisfy requirements of QoS. Therefore, OLT gets several sets of bandwidth requirement for receiving a REPORT message from one ONU. According to the bandwidth computed from inter-scheduling algorithm, OLT determines a appropriate allocation from queue sets in REPORT message. ONU gets granted window size based on REPORT which are send by ONU in previous cycle, so we call it active intra-scheduling algorithm.
Using active intra-scheduling also eliminate unused slotted remainder (USR), besides satisfying requirement of QoS and reasonable packet delay for different class. ONU can’t completely make use of upstream bandwidth which are given by OLT because ethernet packet is not divided, and it will cause USR which is from 0 to 1517bytes.
ONUs contrast granted window size in GATE message with REPORT which was sent to OLT, and it will find a queue set in which sum of queue report equal to granted window size. ONUs transmit packet according to the queue set which is found by contrasting in priority order. For this reason, ONU can completely eliminate USR.
We use the inter-scheduling which is mentioned in other reference and combine it with active intra-scheduling for enhancing throughput, satifying requirement of QoS, and reasonable packet delay.
Finally, we compare the proposed scheme with EAF scheduling in our simulation results. We simulate USR, throughput, packet delay for uniform and non-uniform load to prove our scheme better.
|
author2 |
Chuan-ching Sue |
author_facet |
Chuan-ching Sue Hung-Yuan Chen 陳鴻源 |
author |
Hung-Yuan Chen 陳鴻源 |
spellingShingle |
Hung-Yuan Chen 陳鴻源 Active Scheduling Algorithm for Intra-ONU Dynamic Bandwidth Allocation in Ethernet Passive Optical Networks |
author_sort |
Hung-Yuan Chen |
title |
Active Scheduling Algorithm for Intra-ONU Dynamic Bandwidth Allocation in Ethernet Passive Optical Networks |
title_short |
Active Scheduling Algorithm for Intra-ONU Dynamic Bandwidth Allocation in Ethernet Passive Optical Networks |
title_full |
Active Scheduling Algorithm for Intra-ONU Dynamic Bandwidth Allocation in Ethernet Passive Optical Networks |
title_fullStr |
Active Scheduling Algorithm for Intra-ONU Dynamic Bandwidth Allocation in Ethernet Passive Optical Networks |
title_full_unstemmed |
Active Scheduling Algorithm for Intra-ONU Dynamic Bandwidth Allocation in Ethernet Passive Optical Networks |
title_sort |
active scheduling algorithm for intra-onu dynamic bandwidth allocation in ethernet passive optical networks |
publishDate |
2008 |
url |
http://ndltd.ncl.edu.tw/handle/19719649442951294115 |
work_keys_str_mv |
AT hungyuanchen activeschedulingalgorithmforintraonudynamicbandwidthallocationinethernetpassiveopticalnetworks AT chénhóngyuán activeschedulingalgorithmforintraonudynamicbandwidthallocationinethernetpassiveopticalnetworks AT hungyuanchen zàiyǐtàibèidòngguāngxiānwǎnglùzhōngyǐzhǔdòngshìpáichéngyǎnsuànfǎdádàoguāngxiānwǎnglùdānyuánnèibùdòngtàipínkuānfēnpèi AT chénhóngyuán zàiyǐtàibèidòngguāngxiānwǎnglùzhōngyǐzhǔdòngshìpáichéngyǎnsuànfǎdádàoguāngxiānwǎnglùdānyuánnèibùdòngtàipínkuānfēnpèi |
_version_ |
1718133984980893696 |