Optimal Utilization with Fairness and Priority Considerations in Optical Networks Based on Light Trail

Z. Le, Bisheng Quan, Ming Zhang, Arun Kumar Somani, D. Lastine, K. Balakrishnan
{"title":"Optimal Utilization with Fairness and Priority Considerations in Optical Networks Based on Light Trail","authors":"Z. Le, Bisheng Quan, Ming Zhang, Arun Kumar Somani, D. Lastine, K. Balakrishnan","doi":"10.1109/SOPO.2009.5230251","DOIUrl":null,"url":null,"abstract":"In this paper we consider unidirectional optical busses that originate at a convener node and end at a terminator node. Nodes can transmit only to nodes that are farther downstream, hence the convener node can not receive any bus traffic and the terminator node can not place any information on the bus. The bus supports both data traffic which is bursty as well as connections between source to destination pairs. We assume there exists a control channel outside of the bus that lets the terminator node send information to the convener node. We propose a cycle based protocol where in each cycle all nodes send their requirement to the terminator node which then sends out estimated requirements via the convener node. Then in a distributed computation individual nodes compute their utilization for that cycle based on their priority while paying attention to fair allocation for itself and other nodes. In the process a node may adjust its priority above or below the predicted value. Using simulation we evaluated of the capacity fairness and preemption fairness per node under the constraint of maximum access delay. In order to solve this problem, in this paper we consider unidirectional optical busses that originate at a convener node and end at a terminator node. Nodes can transmit only to nodes that are farther downstream, hence the convener node can not receive any bus traffic and the terminator node can not place any information on the bus. The bus supports both data traffic which is bursty as well as connections between source to destination pairs. We assume there exists a control channel outside of the bus that lets the terminator node send information to the convener node. For this type of network we propose a cycle based protocol where in each cycle all nodes send their requirement to the terminator node which then sends out estimated requirements via the convener node. Then in a distributed computation individual nodes compute their utilization for that cycle based on their priority while paying attention to fair allocation for itself and other nodes. In the process a node may adjust its priority above or below the predicted value. Using simulation we evaluated of the capacity fairness and preemption fairness per node under the constraint of maximum access delay.","PeriodicalId":6416,"journal":{"name":"2009 Symposium on Photonics and Optoelectronics","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2009-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2009 Symposium on Photonics and Optoelectronics","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/SOPO.2009.5230251","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1

Abstract

In this paper we consider unidirectional optical busses that originate at a convener node and end at a terminator node. Nodes can transmit only to nodes that are farther downstream, hence the convener node can not receive any bus traffic and the terminator node can not place any information on the bus. The bus supports both data traffic which is bursty as well as connections between source to destination pairs. We assume there exists a control channel outside of the bus that lets the terminator node send information to the convener node. We propose a cycle based protocol where in each cycle all nodes send their requirement to the terminator node which then sends out estimated requirements via the convener node. Then in a distributed computation individual nodes compute their utilization for that cycle based on their priority while paying attention to fair allocation for itself and other nodes. In the process a node may adjust its priority above or below the predicted value. Using simulation we evaluated of the capacity fairness and preemption fairness per node under the constraint of maximum access delay. In order to solve this problem, in this paper we consider unidirectional optical busses that originate at a convener node and end at a terminator node. Nodes can transmit only to nodes that are farther downstream, hence the convener node can not receive any bus traffic and the terminator node can not place any information on the bus. The bus supports both data traffic which is bursty as well as connections between source to destination pairs. We assume there exists a control channel outside of the bus that lets the terminator node send information to the convener node. For this type of network we propose a cycle based protocol where in each cycle all nodes send their requirement to the terminator node which then sends out estimated requirements via the convener node. Then in a distributed computation individual nodes compute their utilization for that cycle based on their priority while paying attention to fair allocation for itself and other nodes. In the process a node may adjust its priority above or below the predicted value. Using simulation we evaluated of the capacity fairness and preemption fairness per node under the constraint of maximum access delay.
基于光迹的光网络公平性和优先级优化利用
在本文中,我们考虑单向光总线起源于一个召集节点和结束于一个终止节点。节点只能向更下游的节点传输,因此召集人节点不能接收任何总线流量,终止人节点不能在总线上放置任何信息。总线既支持突发的数据流量,也支持源到目的对之间的连接。我们假设总线外存在一个控制通道,该通道允许终止节点向召集节点发送信息。我们提出了一个基于循环的协议,在每个循环中,所有节点将它们的需求发送给终止节点,然后终止节点通过召集人节点发送估计的需求。然后在分布式计算中,各个节点根据其优先级计算其在该周期内的利用率,同时注意自己和其他节点的公平分配。在此过程中,节点可以调整其优先级高于或低于预测值。通过仿真,我们评估了在最大访问延迟约束下每个节点的容量公平性和抢占公平性。为了解决这个问题,本文考虑了单向光总线,它起源于一个召集节点,结束于一个终止节点。节点只能向更下游的节点传输,因此召集人节点不能接收任何总线流量,终止人节点不能在总线上放置任何信息。总线既支持突发的数据流量,也支持源到目的对之间的连接。我们假设总线外存在一个控制通道,该通道允许终止节点向召集节点发送信息。对于这种类型的网络,我们提出了一个基于循环的协议,在每个循环中,所有节点将它们的需求发送给终止节点,然后终止节点通过召集节点发送估计的需求。然后在分布式计算中,各个节点根据其优先级计算其在该周期内的利用率,同时注意自己和其他节点的公平分配。在此过程中,节点可以调整其优先级高于或低于预测值。通过仿真,我们评估了在最大访问延迟约束下每个节点的容量公平性和抢占公平性。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信