- 學位論文
基於Chord演算法與小世界模型之P2P網路搜尋之研究
A P2P Search System Based on Chord Algorithm and Small World Model
摘要
近年來，越來越多的使用者被點對點軟體的便利性所吸引，目前已有許多不同種類的點對點架構。早期的非結構式P2P系統如Gnutella和Napster是以氾濫(Flooding)的方式來傳送查詢訊息，造成網路嚴重負載，而且無法保證在一定的傳送次數後可以找到查詢的目標。因此，近年來學者提出有結構性的點對點架構如CAN、Chord以及Pastry等新的架構解決了以上的缺點。然而大部分的架構都著重於資料的查詢或如何維護與其他用戶之間的溝通，往往沒有考量到使用者查詢時的興趣特性，而造成搜尋資料的時間冗長、網路壅塞及系統效能低落。 本論文基於小世界網路模型(Small-world network model)具體描述了複雜網路的特性、以及使用者興趣的關係。所提出的研究是修改自既有的點對點架構Chord，加上每個使用者的儲存器(Cache)，試圖縮短搜尋資源的時間與減少網路擁塞的問題。最後實驗顯示本論文所提出的P2P架構可以有效率的提升查詢速度，以及減少網路資源的浪費。
並列摘要
In recent years, an increasing number of users were attracted to peer-to-peer (P2P) software’s convenience. There are many different kinds of P2P architectures at present. Early unstructured P2P systems, such as Napster and Gnutella, use flooding way to lookup query messages. Unstructured P2P systems make heavy network loading. It also cannot guarantee that the target will be found. Therefore, many researchers have put forward structural P2P architectures as CAN, Chord, Pastry and other architectures to overcome the above shortcomings. Most of them only focus on how to search data or how to maintain neighbor relations but without consideration of users query properties. Unfortunately, it may result in low performance due to heavy overhead. In order to solve these problems, we propose a new P2P architecture referring to a small-world model which describes the complex network properties and relationships between users interest. The proposed architecture is based on Chord, and use cache in each peer’s storage. We pay more attention to shorten the time to search for resources and reduce the network overhead. Experimental results show that our approach can improve the efficiency of query and thus reduce the network resources consumption.