【作者】 吴尉；

【导师】 陈庆章；

【作者基本信息】 浙江工业大学 ， 计算机应用技术， 2007， 硕士

【摘要】 目前,随着无线网络的发展和个人移动终端的日新月异,人们已经能够并且习惯通过无线网络随时随地地获取信息。同时,对于移动终端之间的资源共享和数据传送也有了越来越多的需求。因此,无线环境下的P2P研究日益受到人们的重视,成为研究的热点。无线环境和P2P的结合由此产生了一个新的研究领域:MP2P。资源定位是MP2P的一个核心问题。一个高效的资源定位机制对于一个成功的MP2P系统而言是必要条件。由于无线环境和移动设备的限制,譬如:高度动态性、有限的带宽、移动设备的可持续时间等,使得无线环境下的P2P与传统的P2P相比有着自己的特点。因此,针对它们的资源定位策略也与通常的有线P2P不同。本文的工作包括:(1)给出一个资源命中率高、命中时延短的模型。在新模型中,采用类DHT机制和区域划分的基于内容的定位方式,有利于进行快速的资源定位。(2)引入了超节点改善资源定位时的命中效率,避免了无超节点时进行洪泛式搜索的性能低下。(3)现有模型并没有仔细考虑网络中节点分布不平衡的情况,本文给出了针对这一情况的解决策略,改善了此情况下的资源定位性能。(4)针对诸多模型中存在的节点振荡,即节点在临界区域频繁反复来回导致资源大量消耗的情况,提出了解决方案。(5)文中我们还将资源定位与资源获取结合起来考虑,即当同一内容有多个提供者时,提供了择优选择节点的算法。并证明该算法的应用对提高传输的稳定性、容错性有帮助。(6)本文给出了新模型的设计方案和实验细节,并在NS2平台上进行模拟验证。通过对比实验,证明新模型与其它模型相比,弥补了它们的一些不足和缺失,同时保证了资源定位的命中率和较短时延,改善了资源获取时的稳定性,具有进一步的研究价值。更多还原

【Abstract】 Nowadays, with the development of both wireless network and personal mobile device, people are accustomed to access information via wireless network from anywhere at anytime. Also, people more and more want to share and transfer the data between their mobile devices. And more and more people are interested in applying P2P under wireless network. There comes out a new concept called MP2P.Resource locating is a key problem of an efficient MP2P system. But with the restraints of wireless network and mobile device, such as highly dynamic topology, low communication bandwidth and limited battery capacity of the device, the strategy of MP2P’s resource locating is different from normal ones.The main work of this paper is as follows.(1)We give a high resource hitting rate and short delay model. It uses the DHT and region partition as our basic resource locating method for quick searching.(2)We introduce super peer to improve efficiency of resource hitting and avoid the poor situation when using flooding request.(3)As existed model didn’t consider the situations of imbalanced distribute very much, we give a strategy to improve the efficiency when the problem occurs.(4) We give the solutions to deal with the ’trashing’ situation existed in most models, which means peers move frequently between two regions’ borderline. When this situation occurs, most of this network’s cost will be expended in dealing with the state change.(5)We consider resource locating and resource retrieval together. When the content has more than one provider, we use the provider selection algorithm. And we prove that it improves the stability and fault-tolerance of data transfers.(6)We present the design and experiment of the new model. Then we simulate and evaluate it on the NS2 platform. From the comparison results of our experiment, we have proved that the new model performs better than some models. And at the same time, the new model ensures the high hitting rate and short delay. Also, it improves the stability of transfers. So the new model worth deeply researching.更多还原