基于P2P的分布式智能UDDI模型研究与设计

发布时间：2018-03-17 13:19

本文选题：智能UDDI模型　切入点：路由算法　出处：《华北电力大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着Internet在各个领域的应用越来越普及，应用种类及数量的迅速增长，人们急切需要一种可以在Internet上实现服务或应用的跨平台、独立及松耦合化的中介技术。此时，Web服务技术应运而生。通过UDDI注册中心的中介功能，将网络上的Web服务有效的整合在一起，供服务请求者使用，但是传统的UDDI发现机制存在网络上的、功能上的性能瓶颈。为了打破这种限制，本文使用MAS(Multi-AgentSystem)对UDDI注册中心进行了功能扩展，并且使用一种混合的P2P网络结构作为注册中心的网络环境，本文主要研究工作如下。首先，Chord作为结构化对等网络中的一种协议类型，主要用于网络中的资源定位，其路由算法的高效性关乎整个网络的服务性能，因此路由算法的查询效率是衡量一个网络好坏的关键指标。研究发现，当网络中存在的节点数远小于网络大小时，每个节点的路由表中会包含多条冗余路由信息，并不利于对资源的查询和定位。因此本文提出了一种基于指针表(fingertable)的Chord路由改进算法：CA-Chord(ContinuousArcChord)路由算法，并通过仿真实验证实，该算法可以消除指针表中的冗余信息，将冗余信息替换为网络逻辑位置相互邻接的一段弧上所包含的节点的路由信息，由此减少了平均查询步长，从而极大的提高了查询效率。其次，结合MAS技术给出了一种智能UDDI模型，将Agent代理封装在UDDI注册中心中，使UDDI模型得到了进一步的优化，，相比传统的UDDI模型，更加自动化和智能化，并通过“多Agent系统工程（MaSE）”方法设计并初步实现了该模型的主要功能。最后，通过使用领域本体、WordNet、语义相似度计算、层次聚类、Qos本体等方法给出了一种基于智能UDDI模型的Web服务多级匹配方法。首先将OWL-S描述的Web服务文档进行WordNet语义扩展，然后将扩展后的文档与领域本体中的概念进行语义相似度匹配，用匹配后的概念替换描述文档中的关键字。服务数量达到某一阈值时，对所存储的服务进行层次聚类。对处理过的服务文档进行类别确定，分类存储，当有服务请求产生时，将该请求与同一类别中的服务进行功能性IO匹配、Qos匹配，最终得到最佳服务。经过实验验证，本方法相比一般基于语义的发现方法，在服务查准率和查全率上均有明显提高。
[Abstract]:With the increasing popularity of Internet applications in various fields and the rapid growth of the types and the number of applications, people urgently need a cross-platform that can implement services or applications on Internet. Independent and loosely coupled mediation technology. At this point, web services technology emerges as the times require. Through the mediation function of the UDDI registry, the Web services on the network can be effectively integrated together for service requesters to use. But the traditional UDDI discovery mechanism has the performance bottleneck on the network. In order to break this limitation, this paper extends the function of the UDDI registry by using MAS Multi-Agent system. And a hybrid P2P network structure is used as the network environment of registration center. The main research work of this paper is as follows. As a kind of protocol type in structured peer-to-peer network, chord is mainly used to locate the resource in the network. The efficiency of the routing algorithm is related to the service performance of the whole network. Therefore, the query efficiency of routing algorithm is the key index to measure the quality of a network. It is found that when the number of nodes in the network is much smaller than the network size, the routing table of each node will contain multiple redundant routing information. Therefore, this paper proposes an improved Chord routing algorithm based on pointer table: CA-Chord-continuous Arc Chord-routing algorithm. The simulation results show that the algorithm can eliminate redundant information in pointer table. The redundant information is replaced by the routing information of nodes contained on an arc in which the logical positions of the network are adjacent to each other, thus reducing the average query step size and greatly improving the query efficiency. Secondly, an intelligent UDDI model based on MAS technology is given. The Agent agent is encapsulated in the UDDI registry, and the UDDI model is further optimized. Compared with the traditional UDDI model, it is more automatic and intelligent. The main functions of the model are designed and implemented by the method of "multiple Agent system Engineering Maze". Finally, a multi-level matching method of Web services based on intelligent UDDI model is proposed by using domain ontology, semantic similarity calculation, hierarchical clustering and QoS ontology. Firstly, the WordNet semantic extension of Web service documents described by OWL-S is introduced. Then the extended document is matched with the concepts in the domain ontology, and the keywords in the description document are replaced by the matched concepts. When the number of services reaches a certain threshold, Hierarchical clustering of stored services. The processed service documents are classified and stored. When a service request is generated, the request is matched with the service in the same category by functional IO matching / QoS matching. Finally, the best service is obtained. The experimental results show that the service precision and recall rate of this method are significantly higher than that of the semantic discovery method.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02

【参考文献】