704 IEEE COMMUNICATIONS LETTERS, VOL. 22, NO. 4, APRIL 2018
SCDN: A Novel Software-Driven CDN for Better Content Pricing and Caching
Jie Duan, Yuan Xing , Ruilin Tian, Guofeng Zhao, Shuai Zeng, Yuanni Liu, and Chuan Xu
Abstract—In conventional content delivery networks (CDNs),
where the content providers (CPs) are lack of the topology
information, it is difficult for them to determine the content
caching and pricing scheme. Furthermore, the small CPs cannot
maintain their own cache servers as it is costly. To solve these
problems, we propose a novel CDN architecture named software-
driven CDN (SCDN) by leveraging the benefit provided by
the software defined networks. In SCDN, the infrastructure
provider (IP) hosts the cache servers at the back of the digital
subscriber line access multiplexer, and it takes over the tasks to
determine the content pricing and caching. By adopting a suitable
revenue sharing scheme, SCDN achieves a win-win situation
between IP and CPs. Simulation results show that with SCDN,
the CPs can get about 25% more revenue than they can get in
conventional CDNs.
Index Terms— SCDN, content pricing, caching, revenue.
I. INTRODUCTION
N
OWADAYS, more and more people get information from
the Internet, entertain themselves with the online video
and audio streaming. All of these result in the exponential
growth of content-based services and applications. It is esti-
mated that the video traffic will be 82 percent of all consumer
Internet traffic by 2021 [1]. Accordingly, it is important for
content providers (CPs) to offer good content delivery services.
To this end, some researchers proposed a framework
named Content Delivery Networks (CDNs) [2]. In CDNs,
the CPs host many cache servers, whenever a customer
requests some content, the corresponding CP let the most
suitable cache server (usually the closest one) to respond
to the request. This is a way to enhance the Quality of
Services (QoS) of the customers, however, it is not the best
solution as it has following drawbacks: 1) CPs do not have the
topology information, which is hosted by the Infrastructure
Provider (IP), and hence it is difficult for the CPs to derive
the best decision on which contents to be cached; 2) it is
difficult to estimate which cache server is the best one that
should respond to the customer’s request, since the closest
one is not always the quickest one that responds to the
request, as the content should be delivered cross the Internet;
3) for small CPs, it is expensive for them to operate their
own cache servers, which limits the application of CDNs.
There are some existing works that are focusing on how
to optimize the content caching and delivery in the CDNs.
Manuscript received January 6, 2018; accepted February 4, 2018. Date
of publication February 8, 2018; date of current version April 7, 2018.
This work was supported by the National Natural Science Foundation of
China under Grant Nos. 61701058, 61501075, 61671130 and 61671124; the
Foundation for Sci & Tech Research Project of Chongqing under Grant
No. cstc2016jcyjA0560. The associate editor coordinating the review of this
letter and approving it for publication was G. Zheng. (Corresponding author:
Guofeng Zhao.)
The authors are with the School of Communication and Informa-
tion Engineering, Chongqing University of Posts and Telecommunications,
Chongqing 400065, China (e-mail: duanjie@cqupt.edu.cn; xingystudy@
foxmail.com; trlwork@foxmail.com; zhaogf@cqupt.edu.cn; zengshuai@
cqupt.edu.cn; liuyn@cqupt.edu.cn; xuchuan@cqupt.edu.cn).
Digital Object Identifier 10.1109/LCOMM.2018.2803808
As one of the representative works, the authors in [3] assume
the CPs host cache servers at the access network in order
to shorten the latency that customers experience and mitigate
the workload in the backbone networks. However, hosting
cache server by each CP is very costly for the small CPs.
Furthermore, they did not consider how to price the content
so as to maximize the revenue. Reference [4] also discusses
how to cache and deliver content based on the cache server
location and routing policy. This work assumes the network
topology is known to the CPs, however, this is not the case
in most of the networks. Cloud-Oriented Content Delivery
Network (CCDN) [5] improves the conventional CDNs by
introducing the Cloud concept. It treats the CDN as multiple
cloud and leverages the technologies in cloud computing
to optimize the content delivery. Similar to other related
works, CCDN does not consider the pricing scheme to opti-
mize the revenue. The CDN-as-a-Virtual-Network-Function
approach [6] is proposed by using the distributed network
functions virtualization platform at the edge of ISP’s network
to solve the ISP-CDN collaboration revenue problem. How-
ever, only the pricing part is considered.
Recently, Software Defined Networks (SDN) [7] is proposed
to enhance the functionality of network plane. Based on the
most famous SDN protocol, OpenFlow, the IP can record
the content request information on the network switches.
In addition, it has the entire network topology information.
Motivated by this fact, we propose a novel CDN architecture
named Software-driven CDN (SCDN). In SCDN, the IP hosts
all the cache servers behind the Digital Subscriber Line
Access Multiplexer (DSLAM) [8], which are used to cache
the contents for the customers in the same area.
In SCDN, we suggest the IP taking over the permission
of pricing and making the cache decision. In this way, not
only the customers can experience a better QoS, but also
the IP can better optimize the pricing and caching schemes.
By leveraging a suitable revenue sharing scheme, both the IP
and CPs can get more revenue with SCDN than that they are
in the conventional CDNs. Even if the CPs would not like
to release the permission of pricing and caching decision to
the IP, they can rent the cache capacity from the IP and offer
a better QoS to the customers and earn more revenue.
The remaining part of this letter is organized as follows.
We introduce the SCDN and discuss how to make the con-
tent pricing and caching decisions in SCDN in Section II,
followed by extensive performance evaluations in Section III.
In Section IV, we conclude this letter.
II. A
RCHITECTURE AND ANALYSIS
To clarify presentation, we first introduce the SCDN archi-
tecture detailedly in Section II-A. Based on the SCDN,
in Section II-B, we theoretically analyze how the IP caches the
content and set the price of the content requested by customers.
Then, we present a way for the IP to share the revenue in
Section II-C. At last, in Section II-D, we discuss the case that
1558-2558 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
资源评论
weixin_38659955
- 粉丝: 4
- 资源: 915
最新资源
- 两 点 间 温 度 控 制
- 基于Matlab的齿轮非线性动力学参数分析：状态变量随参数变化分岔图及周期混沌状态可视化,基于matlab的齿轮非线性动力学，绘出系统状态变量随参数变化分岔图，绘图参数对应的系统各周期及混沌状态的时间
- 基于贝叶斯优化长短期记忆网络(bayes-LSTM)分类预测，matlab代码，要求2019及以上版本 多特征输入单输出的二
- 双向DC-DC变换器（Buck-Boost转换器）仿真：直流电压源对蓄电池的恒流恒压充电与放电模拟（MATLAB Simulink实现）,双向DC DC变器 buck-boost变器仿真 输入侧为直流
- 基于长短期记忆网络(LSTM)的时间序列预测 要求2018b及以上版本，matlab代码 评价指标包括:R2、MAE、MSE
- 基于Matlab的弧齿锥齿轮啮合轨迹分析程序：输出齿轮啮合轨迹及传递误差模拟结果,基于matlab的用于分析弧齿锥齿轮啮合轨迹的程序，输出齿轮啮合轨迹及传递误差 程序已调通，可直接运行 ,核心
- 基于MATLAB的齿轮-轴-轴承系统含间隙非线性动力学建模与动态特性分析,基于matlab的齿轮-轴-轴承系统的含间隙非线性动力学模型，根据牛顿第二定律，建立齿轮系统啮合的非线性动力学方程，同时也主要
- 基于长短期记忆网络LSTM多变量时间序列预测，长短期记忆网络(LSTM)多维时间序列预测，MATLAB代码 评价指标包括:R
- 基于Halcon图像处理的深度示例：二十余种工具集成应用，流程图编写任务执行无忧，完美源码附详细说明文档,基于halcon的图像处理示例，包含二十多个工具，采用流程图方式编写任务，源码可以完美执行，内
- linux常用命令大全.txt
- linux常用命令大全.txt
- 基于贝叶斯优化双向长短期记忆网络(bayes-BILSTM)分类预测，matlab代码，要求2019及以上版本 多特征输入单
- 基于Winform开发的拖拽式运动控制框架：源码展示，未封装，适用于VS2022环境,拖拽式运动控制框架，基于winform开发，除了第三方库都是源码，没封装，VS2022 ,核心关键词：拖拽式运
- 基于QT流程图编辑器拖拽模块源码的VisionMaster风格模仿实现,基于QT流程图编辑器拖拽模块源码，模仿VisionMaster ,核心关键词：QT流程图编辑器; 拖拽模块源码; 模仿Vis
- 基于贝叶斯优化双向长短期记忆网络时间序列预测，bayes-BILSTM时间序列预测，要求2019及以上版本 评价指标包括:R
- 光储一体化机构网型控制仿真模型：探究虚拟同步机控制方式下的光伏与储能系统协同运行机制,光储一体机构网型控制仿真模型 光伏和储能通过各自DC DC变器输出端与DC AC变器的直流侧汇集在一起 DC
资源上传下载、课程学习等过程中有任何疑问或建议，欢迎提出宝贵意见哦~我们会及时处理！
点击此处反馈
