【摘要】 首先从混合式P2P网络拓扑结构出发,结合DHT思想,提出了基于DHT的层次化P2P网络模型。其次根据在文档集巨大的情况下,用户提交的查询不可能"面面俱到",实际用来回答查询的文档仅仅是文档集中很小的一部分这一思想,在层次化P2P模型的超级节点中建立了分布式缓存,运用分布式索引与缓存技术,提出一种新的方法来解决多项查询问题。即由多项查询中的某个关键字key,根据hash函数定位到负责该key的超级节点,查询该节点上的分布式索引得到缓存具体存储位置,最终将结果返回给用户,如若缓存中没有所要查询的内容,则广播该查询,同时根据系统中的历史广播查询信息来计算某个待选缓存项的利益值,利益最大的待选项加入缓存。一般针对多项查询的泛洪算法往往会造成巨大的网络信息量,提出的方法牺牲了超级节点上一小部分的存储力,缓解了多项查询造成的网络拥挤现象。同时,基于DHT的层次化P2P模型也具有很好的稳定性,不会因为大量节点的动态加入或者退出而无法进行多项查询。更多还原

【Abstract】 Proceeding from hybrid P2P network topology structure and combining the DHT idea,a hierarchical P2P network model based on Distributed Hash Table was proposed in this paper at first.If the document sets are tremendous,it is almost impossible that users’ queries can cover every document in the sets,in fact,the documents used to answer queries are only a small parts in the document set.According to this thinking,then we established distributed cache in the super nodes of hierarchical P2P network,a new approach to solve the problem of multi-term queries is given by using the distributed indexing and caching technology.In it,a certain "key" in multi-term queries would be located to the super node in charge of this "key" according to hash function and then search the distributed index in this node to get specific location of the cache,finally the queries from customers are answered.If there is no answer in the cache,then the queries will be broadcasted,meanwhile the historical broadcast queried information in the system is taken as the basis to calculate the interesting value of a certain cache term to be elected,the one with largest value will be added to the cache.Flooding algorithm as the normal technique for processing multi-term queries in P2P environments usually will generate enormous searching trafficin the network.Our method can release network congestion caused by multi-term queries at the cost of losing small storage capability in super nodes.At the same time,hierarchical P2P system based on DHT also has good stability,multi-term queries can be issued even when a large number of nodes join in or quit out of the system.更多还原