P2P網(wǎng)絡(luò)拓撲結(jié)構(gòu)在數(shù)字視頻中的應(yīng)用
3.2萬字 60頁
碩士論文



摘要
P2P（Peer to Peer）技術(shù)在分布式系統(tǒng)和資源共享應(yīng)用領(lǐng)域中得到迅速的發(fā)展，在視頻應(yīng)用領(lǐng)域，近幾年國內(nèi)外涌現(xiàn)出大量的使用P2P技術(shù)的產(chǎn)品。當前非結(jié)構(gòu)化P2P網(wǎng)絡(luò)主要使用集中式查找、洪泛式查找和混合式查找，隨著網(wǎng)絡(luò)規(guī)模的擴大，非結(jié)構(gòu)化P2P網(wǎng)絡(luò)暴露出擴展性能較差的問題。本文針對P2P視頻系統(tǒng)中的一些基本問題進行了研究和實驗，包括：節(jié)點的路由和查找算法、數(shù)據(jù)調(diào)度和節(jié)點選擇策略、網(wǎng)絡(luò)通訊等。本文的主要研究工作和貢獻如下：
（1）節(jié)點的路由和查找算法。查找的效率一直是分布式Hash表一個備受關(guān)注的指標，在分布式哈希表中，比較著名的協(xié)議包括Chord、Pastry、Tapestry和CAN（Content Addressable Network）等。本文結(jié)合具體的應(yīng)用系統(tǒng)的特點和需要修改或者精簡了路由查找算法，使它更適合P2P視頻點播的應(yīng)用。
（2）數(shù)據(jù)調(diào)度和節(jié)點選擇策略。提出了節(jié)點學習歷史經(jīng)驗的數(shù)據(jù)調(diào)度方法和超時數(shù)據(jù)的緊急調(diào)度方法。為了充分利用節(jié)點的資源能力、最大程度地減小中心服務(wù)器的壓力，設(shè)計并實現(xiàn)了節(jié)點學習歷史經(jīng)驗的自適應(yīng)的數(shù)據(jù)調(diào)度和節(jié)點選擇算法；另一方面，為了保證數(shù)據(jù)在規(guī)定的時間內(nèi)到達，設(shè)計并實現(xiàn)了一種超時數(shù)據(jù)的緊急時刻調(diào)度算法。
（3）網(wǎng)絡(luò)通訊。用戶的網(wǎng)絡(luò)環(huán)境多種多樣，有些位于公網(wǎng)，有些位于各種各樣的網(wǎng)關(guān)之后，為了使所有用戶平等地使用系統(tǒng)的服務(wù)，設(shè)計并實現(xiàn)了UDP（User Datagram Protocol）打洞方案解決了部分網(wǎng)關(guān)之后的用戶的通訊問題，使它們達到和公網(wǎng)上的用戶平等的地位。
實驗證明，本文提出的數(shù)據(jù)調(diào)度策略和網(wǎng)絡(luò)通訊方法都是有效的，DHT在較小規(guī)模的實驗環(huán)境中可行，目前還沒有推廣到大規(guī)模的網(wǎng)絡(luò)環(huán)境中。

關(guān)鍵詞：非結(jié)構(gòu)化，結(jié)構(gòu)化，分布式哈希，數(shù)據(jù)調(diào)度，穿越網(wǎng)關(guān)



Abstract
P2P(Peer to Peer) technology has been developing rapidly in fields of distributed system and resource sharing system. VOD(video on demand) based P2P technology becomes one of the hotspots. More and more software and products developed recent years. Centralized and flooding search are widely used in current non-structural P2P system. With the growth of the scales of P2P users, the extension capability becomes weaker. This paper pays attention to several basic issues and design in P2P VOD system, including routing and searching algorithm, choosing sending peers, communication through NAT(Network Address Translator) and so on. The main issues and contributions of this paper are as following:

The first is routing and searching algorithm. Speed is the most important question in searching using DHT(Distributed Hash Table), including some famous protocol such as Chord, CAN(Content Addressable Network), and Pastry. This paper modifies these protocols to adapt for the VOD requirement.

Second, data and sending node selection is related to the quality supplied to users. On one hand, we try our best to reduce the press of central video server, so I designed and realized one data and sending node selection strategy called self adaptive or study based selection. On another hand, in order to ensure the download time of data, I designed and realized one data and sending node selection strategy called emergent selection.

Not all the users are in Internet, but some users are behind NAT. In order to supply all users with the same usage, this paper designed and realized one method to communicate through NAT equally.

Experiment proves that the data and sending node selection strategy and communication method through NAT suggested in this paper is valid. DHT can act in small scale system, but still has not been used in large scale system.

Key Words: non-structural, structural, DHT(Distributed Hash Table), data and sending node selection, NAT



目 錄
第一章 緒論 1
1.1 P2P技術(shù)簡介 1
1.2 P2P在視頻點播中的應(yīng)用 3
1.3 國內(nèi)P2P視頻發(fā)展現(xiàn)狀 6
1.4 國外P2P視頻發(fā)展情況 8
1.5 P2P視頻以何制勝 9
1.6 面臨的挑戰(zhàn) 9
1.7 本文的主要工作和文章結(jié)構(gòu) 10
第二章 P2P查找算法及改進 12
2.1 P2P網(wǎng)絡(luò)模型 12
2.2 分布式哈希搜索算法 16
2.2.1 Chord協(xié)議 16
2.2.2內(nèi)容訪問網(wǎng)絡(luò)（Content-Addressable Network，CAN） 24
2.2.3 Pastry系統(tǒng) 29
第三章 關(guān)鍵技術(shù)的研究和運用 35
3.1 項目簡介 35
3.2 基于劃分區(qū)域的查找 38
3.3 數(shù)據(jù)調(diào)度策略 40
3.4 復(fù)雜網(wǎng)絡(luò)環(huán)境中的通訊 46
3.4.1 基本概念 46
3.4.2 面臨的困難及其解決的方法 48
3.4.3 實驗 51
第四章 總結(jié)和展望 53
參考文獻 54
致謝 56



參考文獻
[21] 董海韜， 基于對等網(wǎng)絡(luò)流媒體傳輸系統(tǒng)技術(shù)的研究，清華大學， 2005.5
[22] 馬凌霄，基于p2p網(wǎng)絡(luò)的流媒體技術(shù)研究，浙江大學， 2005年
[23] Chao Zhang, Hai Jin, Dafu Deng, Sirui Yang, Quan Yuan and Zuoning Yin, Cluster and Grid Computing Lab. Anysee: Multicast-based Peer-to-Peer Media Streaming Service System, 2005 Asia-Pacific Conference on Communications, Perth, Western Australia, 3 - 5 October 2005.
[24] Xinyan Zhang, Jiangchuan Liu, Bo Li, and Tak-Shing Peter Yum, CoolStreaming/DONet: A Data-driven Overlay Network for Peer-to-Peer Live Media Streaming, 2004
[25] Venkat Padmanabhan, Microsoft Research, Resilient Peer-to-Peer Streaming, 2003.3
[26] Kai Hwang, Internet Security and Cyber Trust for Integrated Web Services and P2P Grid Computing, 2005.12
[27] Meng Zhang, Jian-Guang Luo, Li Zhao, and Shi-Qiang Yang, A Peer-to-Peer Network for Live Media Streaming – Using a Push-Pull Approach, Tsinghua University, 2005
[28] Xiaojun Hei, Chao Liang, Jian Liang, Yong Liu and Keith W. Ross, Insights into PPLive: A Measurement Study of a Large Scale P2P IPTV System