CDN-P2P架构下的文件下载策略的研究与实现

发布时间：2018-05-14 14:56

本文选题：CDN-P2P + 相似性　；参考：《北京邮电大学》2014年硕士论文

【摘要】：本文研究并分析了CDN和P2P技术的优缺点,并对CDN-P2P融合网络的特点进行了研究,针对其下载过程,从P2P网络的节点组织结构,资源搜索策略以及CDN网络中的副本放置策略等三个不同的层次分析、优化CDN-P2P网络,以便更好地结合这两种技术的优势提高文件分发下载的效率和服务质量,提升用户体验。主要有以下几个关键点：第一从用户节点的组织结构考虑,改进了基于相似性的邻居选择算法。将节点对资源类型的偏好相似性和节点对资源的评分相似性结合起来,形成节点整体的兴趣相似性,将兴趣相似性较高的节点组织在一个子网中,使得下载更有针对性。第二从P2P网络的资源定位角度出发,提出了一种基于兴趣差别的查询算法,避免盲目转发查询消息,使得查询更有针对性,提高资源定位的效率,减少下载过程中因查询产生的额外流量。第三从CDN网络的资源放置策略出发,边缘服务器通过分析节点在搜索、下载过程中的数据以及资源的类型信息,定期运行内容预存任务,向中心服务器反馈近期热门类型及其需求数量。中心服务器通过组织数据库文件查询将相应的资源列表返回给边缘服务器,满足其预存需要,以便快速响应节点的文件请求,提高节点用户发现所需求内容以及共享内容的效率,进而提高下载效率。
[Abstract]:In this paper, the advantages and disadvantages of CDN and P2P technology are analyzed, and the characteristics of CDN-P2P fusion network are studied. According to the download process, the node organization structure of P2P network is introduced. In order to improve the efficiency and service quality of file distribution and download and improve the user experience, the resource search strategy and the replica placement strategy in CDN network are analyzed in order to optimize the CDN-P2P network and improve the efficiency and service quality of file distribution and download. There are the following key points: First, considering the organization structure of the user node, the neighbor selection algorithm based on similarity is improved. The similarity of node preference to resource type and the similarity of node score on resource are combined to form the interest similarity of nodes as a whole, and the nodes with high interest similarity are organized into a subnet to make the download more targeted. Second, from the point of view of resource location in P2P network, a query algorithm based on interest difference is proposed, which avoids blindly forwarding query messages, makes the query more targeted and improves the efficiency of resource location. Reduce the additional traffic generated by the query during the download process. Thirdly, starting from the resource placement strategy of CDN network, the edge server runs the task of content pre-storage periodically by analyzing the data in the search and download process and the type information of the resource. Feedback to the central server the number of recent hot types and their requirements. The central server returns the corresponding list of resources to the edge server by organizing database file queries to meet its pre-storage needs in order to respond quickly to file requests from nodes. Improve the efficiency of node users to discover the required content and share content, and then improve the efficiency of downloading.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02

【参考文献】