【作者】 夏军；

【导师】 崔国华；

【作者基本信息】 华中科技大学 ， 信息安全， 2011， 硕士

【摘要】 目前P2P网络的应用越来越广泛,P2P技术已经应用很多的领域。P2P网络具有动态性、可扩展性、自组织、非中心化的特点,使得P2P网络存在很大的安全隐患,使得节点之间缺乏信任,很大程度上限制了P2P网络的进一步发展,因此能否有效可靠的对节点进行行为分析与预测是保障高质量资源共享的重要因素。针对P2P网络特点,在深入研究灰色理论的基础上,将灰色关联投影、灰预测相结合,阐述了基于投影的灰色行为分析模型(GBAP),对对等网络(P2P)中的节点进行行为分析与预测。主要分成两个部分,包括节点的行为分析和节点的行为预测。节点的行为分析主要是将多目标决策的灰色关联投影用于节点的鉴别,计算各个节点关于理想节点的映射值,以此来作为节点性能评判的指标。节点的行为预测主要是对每个节点的映射值进行灰生成,达到极性统一的效果,然后建立灰模型,这里采用平滑型GM(1,1)模型,之后进行灰预测,以此作为节点未来行为预测的指标,最后计算节点的滚动残差,得到节点的可信度,对灰预测模型的性能进行阐述。在采集P2P网络数据的基础上,计算了四个节点相对于理想节点的映射值,并对上述数据进行了主成分分析,将映射值和主成分分析结果进行了对比,同时对节点进行了预测,对四个节点的各方面性能进行了详细分析与比较,结果表明,基于投影的灰色行为分析模型是对P2P网络中节点行为分析与预测的有效方法,该方法对提高P2P网络评价的客观性有着积极的意义,对P2P网络的安全通信奠定了良好的基础。更多还原

【Abstract】 Currently P2P networks used more wide, and P2P technology has been applied in a lot of fields. P2P network is dynamic, expansible, self-organization and non-centralized, which result in a big security problem in the P2P network and the lack of trust between nodes, so that largely impede the further development of P2P networks. Therefore nodes behavior analysis and prediction is an important factor in the sharing of resources.For the features of P2P networks and on the basis of deep study of grey theory, we correlate grey relation projection to grey prediction and propose grey behavior analysis based on projection method or GBAP in short, for the behavior analysis and prediction of nodes in P2P networks. GBAP can be divided into two parts, including the node behavior analysis and the node behavior prediction. The work of node behavior analysis is to apply grey relation projection to the node identification, then calculate projection value of each candidate node to the ideal node, in order to evaluate the performance of node. The node behavior prediction is to apply grey generating on the projection value of each node, in order to unify the goal nature of series, and then generate grey module, here we use SmoGM(1,1) module, after that, we calculate the prediction value for the behavior prediction of node, at last calculate the residual error and get the credibility of the node, the analysis the performance of the grey prediction module.On the base of collect all the data of P2P network, paper calculate projection value of the four candidate node to the ideal node, and apply principal component analysis on the same data, the compared the projection value and the result of principal component analysis, then carry detailed analysis and comparison on the all kinds aspects of four nodes performance. The result show that grey behavior analysis based on projection is an effective method for the nodes behavior analysis and prediction in P2P network, it helps improve the objectivity of P2P networks evaluation and lay a good foundation for secure communications in the P2P network.更多还原