基于兴趣分组的混合P2P网络研究

发布时间：2018-08-12 09:33

【摘要】：随着网络的多样化发展,对等网络(P2P,Peer to Peer)作为一种新型的网络应运而生并迅速发展。分布式P2P网络中的节点(Peer)之间地位是相等的并且他们之间可以直接交互,整个网络的性能理论上是所有节点性能的总和,相对于传统的集中式拓扑网络可扩展性也更好。伴随P2P网络的规模及用户的增多,P2P技术的应用领域也越来越广泛,进而整个环境中的信息量也呈现飞速增长,用户在P2P网络进行搜索、定位及获取资源时也越来越困难。研究P2P网络的信息搜索技术成为解决这一问题的重要手段。个好的搜索技术不但能有效的提高路由效率,增加搜索命中率,还能根据资源的兴趣分布缩短搜索的范围,节约用户的搜索时间,提高工作效率。由此可见,研究如何优化P2P网络的搜索技术这一课题具有很大的价值。现如今围绕P2P搜索技术课题的研究大多是关系结构化P2P网络的,因此,本文在以往研究的基础上,提出一种基于兴趣分组的混合P2P网络IGTAP(Int erest-Group Tapestry model).它以Small World为理论依据,将分组的理念引入到网络中,以节点的兴趣作为分组的依据,将整个P2P网络分为不同的兴趣组。同时引入分层的概念,以Tapestry作为底层拓扑结构,将Tapestry层分为不同的兴趣组,节点根据自己的兴趣加入到不同的兴趣组中。兴趣组内的节点之间的兴趣距离是根据兴趣相似度计算得来的。第二层的树形结构是由邻居映射表伸展成的一个多叉树,每个兴趣组下都有一个多义树,并且每个节点只能加入到一个兴趣组。网络中的搜索策略分为组内搜索和组间搜索。组内搜索是指节点对本兴趣组内资源的搜索；组间搜索是在组内搜索失败或者所要搜索的资源不在本组内时发生的情况。搜索的过程中还引入缓冲池策略,很好的解决了短时间内搜索相同资源而引起的网络资源的浪费。最后通过对网络的分析和模拟实验,结果表明本文方案有效的提高了查询成功率,缩短了路由跳数,同时提高了搜索的精准率。
[Abstract]:With the development of network diversification, peer-to-peer network (P2P) emerges and develops rapidly as a new type of network. Peers in distributed P2P network have equal status and can interact with each other directly. The performance of the whole network is theoretically the sum of all the nodes'performance, which is relative to the traditional centralization. With the increase of the scale of P2P network and the number of users, the application fields of P2P technology are more and more extensive, and the amount of information in the whole environment is also increasing rapidly. It is more and more difficult for users to search, locate and obtain resources in P2P network. A good search technology can not only effectively improve routing efficiency, increase the hit rate of search, but also shorten the search scope according to the interest distribution of resources, save the search time of users and improve work efficiency. Value.
Nowadays, the research on P2P search technology is mostly about relational structured P2P networks. Therefore, based on previous research, this paper proposes a hybrid P2P network IGTAP (Int erest-Group Tapestry model) based on interest grouping. It introduces the concept of grouping into the network based on the theory of Small World and takes the interest of nodes as its basis. As the basis of grouping, the whole P2P network is divided into different interest groups. At the same time, the concept of hierarchy is introduced. Tapestry is used as the underlying topology structure, and Tapestry layer is divided into different interest groups. The nodes join different interest groups according to their own interest. The interest distance between the nodes in interest groups is calculated according to the interest similarity. The tree structure of the second layer is a multi-tree extending from the neighborhood mapping table. Each interest group has a multi-tree and each node can only join one interest group. Cable occurs when the group search fails or the resources to be searched are not in the group. Buffer pool strategy is introduced to solve the waste of network resources caused by searching the same resources in a short time.
Finally, through the network analysis and simulation experiments, the results show that the proposed scheme effectively improves the query success rate, shortens the number of routing hops, and improves the search accuracy.
【学位授予单位】：大连交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02

【相似文献】