本文選題：Kademlia + 活躍節(jié)點　； 參考：《華中科技大學(xué)》2014年博士論文

【摘要】：自2002年Kademlia協(xié)議算法被提出后,基于Kademlia的DHT網(wǎng)絡(luò)研究與應(yīng)用備受關(guān)注,各類系統(tǒng)及應(yīng)用軟件層出不窮。支持該協(xié)議的DHT網(wǎng)絡(luò)文件分享系統(tǒng)常常擁有百萬級的用戶數(shù),使得Kademlia成為實際應(yīng)用最廣泛的文件分享網(wǎng)絡(luò)協(xié)議之一。與其它P2P網(wǎng)絡(luò)系統(tǒng)一樣,節(jié)點動態(tài)參與網(wǎng)絡(luò)的特性影響了Kademlia DHT網(wǎng)絡(luò)的性能。然而在KademliaDHT網(wǎng)絡(luò)中,有些節(jié)點長時間參與網(wǎng)絡(luò),無私的進(jìn)行網(wǎng)絡(luò)文件分享,成為網(wǎng)絡(luò)中的核心節(jié)點,我們稱其為活躍節(jié)點。本文將基于Kademlia DHT網(wǎng)絡(luò)中這些活躍節(jié)點的特性,針對目前研究與應(yīng)用中的幾個核心問題進(jìn)行展開： ·節(jié)點在線行為分析與預(yù)測：在Kademlia DHT網(wǎng)絡(luò)中,節(jié)點的在線情況和參與網(wǎng)絡(luò)的行為直接決定了整個網(wǎng)絡(luò)的性能。所以節(jié)點的在線行為分析與預(yù)測一直是研究的關(guān)注點。 ·內(nèi)容發(fā)布與查詢性能提升：目前各類支持Kademlia DHT網(wǎng)絡(luò)的客戶端軟件主要用于文件分享。所以在Kademlia DHT網(wǎng)絡(luò)的測量研究中,如何利用測量結(jié)果提升內(nèi)容發(fā)布與查詢性能一直備受關(guān)注。現(xiàn)有的Kademlia DHT網(wǎng)絡(luò)雖然支撐了百萬級用戶的實時內(nèi)容發(fā)布與查詢,但還是有部分內(nèi)容發(fā)布后查詢不到,性能還有待提高。 ·基于Kademlia DHT網(wǎng)絡(luò)的衍生服務(wù)：在Kademlia DHT網(wǎng)絡(luò)中,參與節(jié)點把自身擁有的帶寬、存儲、計算等網(wǎng)絡(luò)資源貢獻(xiàn)出來共同組成了現(xiàn)在的文件分享網(wǎng)絡(luò)。如何更好的去利用這些通過Kademlia DHT網(wǎng)絡(luò)集結(jié)的網(wǎng)絡(luò)資源衍生更多的網(wǎng)絡(luò)服務(wù)是一個新的研究方向。 本文針對上述問題,基于活躍節(jié)點對KAD網(wǎng)絡(luò)和Mainline DHT網(wǎng)絡(luò)進(jìn)行了測量分析。本文的研究從以下三個方面展開：(1)Kademlia DHT網(wǎng)絡(luò)中活躍節(jié)點的獲取；(2)基于活躍節(jié)點的內(nèi)容發(fā)布與查詢算法改進(jìn)；(3)基于Kademlia DHT穩(wěn)定節(jié)點的Web緩存系統(tǒng)。這些工作得到國家自科基金No.61301127、No.61371080和No.61370231的支持；以及國家“十二五”科技支撐計劃項目No.2011BAK08B01和No.2011BAK08B02-01的資助。 本文取得的研究成果包括以下幾個方面： 1. Kademlia DHT網(wǎng)絡(luò)中活躍節(jié)點的快速獲取機制：本文對KAD網(wǎng)絡(luò)和Mainline D-HT網(wǎng)絡(luò)進(jìn)行了測量分析,發(fā)現(xiàn)可以使用節(jié)點ID在路由表中的出現(xiàn)頻率來表征節(jié)點在網(wǎng)絡(luò)中的活躍度,并給出了活躍節(jié)點的快速定位獲取算法。通過該算法獲得的活躍節(jié)點較為穩(wěn)定,較長時問持續(xù)參與網(wǎng)絡(luò)。 2.基于活躍節(jié)點的內(nèi)容發(fā)布與查詢算法：本文測量分析了Kademlia DHT網(wǎng)絡(luò)的內(nèi)容發(fā)布與查詢過程,提出了基于活躍節(jié)點的內(nèi)容發(fā)布與查詢算法,通過該算法,能顯著地提高內(nèi)容發(fā)布與查詢的性能。 3.基于KAD穩(wěn)定節(jié)點的Web緩存系統(tǒng)：針對基于Kademlia DHT的Web緩存系統(tǒng)中緩存節(jié)點的穩(wěn)定問題,本文提出了基于活躍節(jié)點的緩存節(jié)點選擇算法,選擇KAD網(wǎng)絡(luò)中較為穩(wěn)定的節(jié)點來擔(dān)當(dāng)Web緩存節(jié)點,從而提高系統(tǒng)性能。
[Abstract]:Since the Kademlia protocol algorithm was proposed in 2002, the research and application of DHT network based on Kademlia have attracted much attention. The DHT network file sharing system which supports this protocol often has millions of users, which makes Kademlia one of the most widely used file sharing protocols in practice. Like other P2P networks, the dynamic participation of nodes affects the performance of Kademlia DHT networks. However, in the KademliaDHT network, some nodes participate in the network for a long time, selfless network file sharing, become the core node in the network, we call it an active node. Based on the characteristics of these active nodes in Kademlia DHT network, this paper aims at several core problems in current research and application. Node online behavior analysis and prediction: in Kademlia DHT network, the performance of the whole network is directly determined by the online behavior of the node and the behavior of the participating network. Therefore, the online behavior analysis and prediction of nodes has been the focus of research. Content publishing and query performance improvement: at present, all kinds of client software supporting Kademlia DHT network are mainly used for file sharing. Therefore, in the measurement research of Kademlia DHT network, how to use the measurement results to improve the performance of content publishing and query has attracted much attention. Although the existing Kademlia DHT network supports the real-time content publishing and querying of millions of users, there are still some contents that can not be queried after publishing, and the performance still needs to be improved. Derivative Services based on Kademlia DHT Network: in Kademlia DHT network, participating nodes contribute their own bandwidth, storage, computing and other network resources to form the present file sharing network. How to make better use of these network resources accumulated through Kademlia DHT network to derive more network services is a new research direction. In this paper, KAD network and Mainline DHT network are measured and analyzed based on active nodes. In this paper, the following three aspects of research are carried out: getting active nodes in Kademlia DHT network) improving the content publishing and query algorithm based on active nodes and improving the Web cache system based on Kademlia DHT stable nodes. The work is supported by the National Science and Technology Foundation No. 61301127 No. 61371080 and No.61370231; and supported by the National 12th Five-Year Plan for Science and Technology support Project No.2011BAK08B01 and No.2011BAK08B02-01. The research results of this paper include the following aspects: 1. The Fast access Mechanism of active nodes in Kademlia DHT Networks: this paper measures and analyzes the KAD network and Mainline D-HT network, and finds that the frequency of node ID appearing in the routing table can be used to characterize the node activity in the network. A fast location acquisition algorithm for active nodes is presented. The active nodes obtained by this algorithm are stable and participate in the network continuously for a long time. 2. Content publishing and querying algorithm based on active nodes: this paper measures and analyzes the content publishing and querying process of Kademlia DHT network, and puts forward the content publishing and querying algorithm based on active nodes. It can significantly improve the performance of content publishing and query. 3. Web Cache system based on KAD stable Node: aiming at the problem of Cache Node Stability in Web Cache system based on Kademlia DHT, this paper proposes an active Node based Cache Node selection algorithm. The more stable nodes in KAD network are selected to act as Web cache nodes to improve system performance.
【學(xué)位授予單位】：華中科技大學(xué)
【學(xué)位級別】：博士
【學(xué)位授予年份】：2014
【分類號】：TP393.02

【參考文獻(xiàn)】

相關(guān)期刊論文 前4條

1 嚴(yán)鶴;劉威;張戈;程文青;;基于活躍節(jié)點的KAD網(wǎng)絡(luò)高效查詢方法[J];計算機科學(xué);2014年12期

2 ;An implementation and optimization for scalable DHT crawler[J];Science China(Information Sciences);2010年04期

3 張宇翔;楊冬;張宏科;;P2P網(wǎng)絡(luò)中Churn問題研究[J];軟件學(xué)報;2009年05期

4 劉祥濤;程學(xué)旗;李洋;陳小軍;白碩;劉悅;;eMule網(wǎng)絡(luò)最佳節(jié)點標(biāo)識[J];軟件學(xué)報;2011年09期

，

本文編號：1851574