论文部分内容阅读
对等网文件共享系统能否成功极大地取决于搜索机制的多样性和扩展性 当前支持分布式Hash表 (DHT)功能的结构化系统 (如CAN)易扩展但不能有效地支持部分匹配的查询 ,而基于扩散的非结构化系统 (如Gnutella)支持多样化查询但不易扩展 提出了一种新的对等网体系结构 基于CAN构造的pGroup介于结构化和非结构化之间 :结点根据内容的类别自组织在一起 ,具有相同类别的结点相互关联构成语义对等网 (semanticpeer to peernetworks ,SPNs) 针对不同的查询行为 ,提出了相应的搜索算法 实验表明 ,pGroup比Gnutella的扩展性好 ,搜索效率高
The success of peer-to-peer file sharing systems greatly depends on the diversity and scalability of search mechanisms. Structured systems (such as CAN) that currently support distributed hash table (DHT) capabilities are extensible but do not effectively support partially-matched queries , Whereas diffusion-based unstructured systems such as Gnutella support diverse queries but are not easily scalable. A new peer-to-peer network architecture is proposed. A pGroup based on CAN construction is between structured and unstructured: The categories of content are self-organizing and the nodes of the same category are interrelated to form semantic peer to peer networks (SPNs). According to different query behaviors, the corresponding search algorithm is proposed. Experiments show that pGroup has better scalability than Gnutella , Search efficiency