【作者】 王彦昆；

【导师】 贾宗璞； 冯文峰；

【作者基本信息】 河南理工大学 ， 计算机应用技术， 2009， 硕士

【摘要】 随着网络技术的发展,流媒体技术也得到了广泛的应用。传统的C/S模式的流媒体系统的资源随着用户的增加很快就被消耗完,成为系统瓶颈。P2P的出现为流媒体的发展提供了新的方向。在P2P模式中媒体内容通过使用客户自己的带宽分发,每个节点具备客户机和服务器功能,作为服务请求节点,在接收网络服务的同时,作为服务提供节点,利用自身剩余的能力为网络中其他节点提供服务,这样就在很大程度上减轻了服务器的负担。但现存的多数P2P流媒体系统用时延或负载平衡等其中的一种或两种属性作为度量值,并没有综合考虑节点可靠性、网络带宽、数据包丢失率及整个算法的维护开销。本文以CollectCast节点管理模块为思想基础,提出了较快速准确、较低资源消耗的网络拓扑发现方法以及基于此的最佳节点选择算法。论文主要贡献有以下四点:一是提出和实现了快速并行面向群组的Traceroute方法,该方法通过深度和广度上的并发大大提高了拓扑树的生成速度;二是使用PingGraph工具测量可用带宽估测和丢包率,减少了网络资源的使用和测量的时间;三是提出和实现了基于拓扑树遍历的带宽瓶颈探测算法,通过3次遍历对节点实际可用带宽的计算,有效的避免的了瓶颈的发生;四是提出和实现了以“综合结点可用性分数”来统一衡量节点综合可用性的新节点评估方法。通过实验可以看出该方法硬件资源消耗较少,网络资源占用较低,节点的可用性通过IPAM表现了出来。更多还原

【Abstract】 With the developing of the network technology, Streaming media technology has been applied extensively. Resource of traditional C/S model streaming media system is to be run out quickly with increasing of user number, become system’s bottleneck. The appearance of P2P provides new technology for streaming media. In the P2P model, contents are distributed by using the bandwidth of the clients themselves, each node with the client and server functionality, as a service request node, at the same time to receive network services, as a service provider node, using its own remainder capacity provide services for the other nodes in the network, to a large extent alleviated the burden on the server. However, most of the existing P2P streaming media system with delay or load balancing, and one or two degrees as the property value did not consider the reliability of nodes, network bandwidth, packet loss rate and the maintenance expenses whole algorithm.The thesis developed more rapid, more accurate but lower consumption methods of network topology inference and the following peer selection, which referred the CollectCast peer management module as their ideological basis. The following four points are the main contribution of the thesis. First, it proposed and achieved a rapid, parallel, and group-oriented Traceroute, which adopt the breadth and depth of complicated greatly increased the rate of topologic tree formation. Second, use PingGraph measure available bandwidth and packet loss rate, which reduce the use of network resource and the measurement of time. Third, it proposed and implemented the bandwidth bottleneck detection algorithm through traversing the network topologic tree; calculate available bandwidth through three times traverse effectively prevent the occurrence of the bottleneck. And forth, it proposed and achieved a new peer assessment method, which included a new scale, Integrated Peer Availability Mark, to measure the peer’s integrated availability. It can be seen from the result of experiment, this method use less resource of hardware and network, IPAM show the availability of peer.更多还原