IPv6环境下基于DHT的P2P媒体共享系统的研究与实现

发布时间：2018-05-23 08:37

本文选题：P2P + DHT　；参考：《北京邮电大学》2014年硕士论文

【摘要】：随着网络规模的扩大和用户数量的急剧增加,使得互联网的流量急剧增加,给网络的带宽造成极大的压力,也加大了服务器的负担,导致服务不稳定。当前C/S架构网络将信息的传输中转和处理都集中在中心服务器上,容易造成网络拥塞和性能瓶颈。于此同时,个人计算机的飞速发展,使得每台计算机相比过去都具有较强的计算和存储能力,而这些性能在大部分情况下没有得到充分的利用。在这种背景下产生了P2P技术,网络中各节点相互平等独立,没有中央服务器,节点与节点之间直接进行信息和数据的传输。一方面将中心服务器上的任务分散到网络中的各个节点,极大的降低了服务器的和网络带宽的压力,同时由于P2P的节点分布特性,使其具有很强的健壮性和可扩展性,避免了单点失效；另一方面也使得个人计算机中未被发掘的性能得到更充分的利用。由于没有中心服务器,资源分布在网络内各个节点中,所以在P2P网络中,最关键的问题就是如何找到快速找到存储资源的节点。第一代的P2P系统为基于目录服务器的架构,虽然资源都存储在节点上,但是需要通过目录服务器记录特定的资源的存放位置,节点需要先与目录服务器通信,得到文件在网络上的存储信息,最后才能与节点直接通信获得资源,在用户增多时,目录服务器同样也会存在瓶颈问题。第二代是非结构化的架构,采用泛洪的方式在网络中查询文件的存储位置,在网络规模变大时,会给网络带来较大的负担,而且查找效率较低。针对以上系统的问题,提出了第三代基于DHT的P2P网络。DHT将文件的索引信息存在各个节点上,取代了目录服务器,然后每个节点利用自身保存的路由信息,通过特定的路由算法就可以快速找到文件的索引信息,最后直接与存储着文件的节点进行通信,目前应用比较广泛的是Kademlia、Chord、Pastry、CAN等算法。目前基于Kademlia的DHT网络还存在许多问题和需要改进的地方。首先是基于Kademlia的DHT实现中还没有基于IPv6网络的,无法在纯IPv6环境下使用DHT功能。其次是DHT中节点的逻辑位置和物理位置完全脱离,使得路由查找时效率不够高。还有就是当前DHT的架构中,由于各个节点完全平等,节点经常从其他网络寻找资源造成过多的跨域流量。针对以上问题通过对Libtorrent函数库进行改造,针对IPv6网络进行修改,同时融合IPv6的特性,将物理位置信息融入到节点之中。通过对DHT进行层次化优化,使得DHT网络具有更好的路由效率。最后利用以上改进,设计并实现了一个IPv6环境下基于DHT的P2P媒体分享系统,检验优化方案的效果。
[Abstract]:With the expansion of the network scale and the sharp increase of the number of users, the traffic of the Internet increases rapidly, which puts great pressure on the bandwidth of the network, increases the burden of the server, and leads to the instability of the service. The current C / S network concentrates the transmission and processing of information on the central server, which can easily lead to network congestion and performance bottlenecks. At the same time, with the rapid development of personal computers, each computer has stronger computing and storage capacity than in the past, and these properties are not fully utilized in most cases. In this context, P2P technology has emerged. Each node in the network is equal and independent, and there is no central server, and the information and data are transmitted directly between the nodes. On the one hand, the task on the central server is distributed to the nodes in the network, which greatly reduces the pressure of the server and the network bandwidth. At the same time, because of the distribution of P2P nodes, it has a strong robustness and scalability. The single-point failure is avoided; on the other hand, the undiscovered performance of personal computer is more fully utilized. Since there is no central server and the resources are distributed among the nodes in the network, the most important problem in P2P network is how to find the node of storage resource quickly. The first generation of P2P system is based on directory server architecture. Although the resources are stored on the node, it is necessary to record the location of the specific resources through the directory server, and the node needs to communicate with the directory server first. Only when the file is stored on the network can it communicate directly with the node to obtain the resource. When the number of users increases, the directory server will also have a bottleneck problem. The second generation is an unstructured architecture. When the network scale becomes larger, it will bring a great burden to the network, and the search efficiency will be low by using flooding method to query the storage location of the files in the network. In order to solve the above problems, the third generation P2P network based on DHT is proposed to store the index information of the file on each node, replacing the directory server, and then each node uses the routing information saved by itself. The index information of files can be found quickly by a specific routing algorithm, and finally it can communicate directly with the nodes that store the files. At present, the algorithms such as Kademlia Chorden Pastrycan and so on are widely used. At present, the DHT network based on Kademlia still has many problems and needs to be improved. First of all, there is no IPv6 network in the DHT implementation based on Kademlia, so it is impossible to use DHT in pure IPv6 environment. Secondly, the logical and physical positions of nodes in DHT are completely detached, which makes the routing lookup inefficient. In the current DHT architecture, due to the equality of all nodes, nodes often seek resources from other networks, resulting in excessive cross-domain traffic. In view of the above problems, the Libtorrent function library is modified, the IPv6 network is modified, and the characteristics of IPv6 are fused, and the physical position information is integrated into the nodes. Through hierarchical optimization of DHT, DHT network has better routing efficiency. Finally, a P2P media sharing system based on DHT under IPv6 environment is designed and implemented by using the above improvements, and the effect of the optimized scheme is tested.
【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】