【作者】 付秋林；

【导师】 李学明；

【作者基本信息】 重庆大学 ， 计算机软件与理论， 2008， 硕士

【摘要】 对等计算(P2P)是近年来研究的热点,其目标是充分利用Internet环境中各种可能的端系统进行大规模的协作计算和资源共享。由于下述问题直接影响了对等计算的可用性和实用性,因此对其研究既有理论意义也有实用价值。P2P计算环境具有开放性以及对等节点的匿名性和自治的特点,由于缺乏对与之交互的节点的可信程度的知识,节点需应对交互过程中可能出现的威胁。引入信任模型量化和评估节点的可信程度,是解决问题的一种方法。本文着重研究这个方面的问题。主要的贡献概括如下1.综述了环境下的信任模型。对P2P环境下信任模型的研究工作做了详细地整理、分类和评析。2.提出了基于节点距离加权推荐的全局信任模型。该模型针对P2P环境下信任模型面临的主要挑战:恶意节点协同伪造信任数据对模型攻击,使得信任评估的结果不真实;信任数据的分布式存储及其安全性。已有的两种全局信任模型建立在信任值高的节点其推荐也更可信这个假设基础上,而在某些类型的恶意节点攻击下,该假设并不成立。因此本模型中,提出使用节点的距离加权其推荐。理论分析和仿真实验表明较已有模型更加适合遏制更广泛类型的恶意节点攻击。给出了一种信任数据的分布式存储方案,提出了加强数据安全性的对信任数据存放节点的匿名性的三个要求。更多还原

【Abstract】 Nowadays Peer-to-Peer(P2P) computing has become a hot research issue. The goal of P2P computing is to pool end-systems in Internet-scale as more as possible to construct large-scale cooperating and resource-sharing environments.Because the following problem directly decide the usability and practicability of P2P system, it has been in dire need of a solution. In decentralized P2P environments, due to the anonymous and autonomous nature of peers, they have to manage the risk involved with the interaction without prior experience and knowledge about each other’s reputation. Trust models can be used to quantify and to evaluate the trustworthiness of peers so as to reduce the risk.The dissertation focuses on this problem, and makes these contributions:1. A comprehensive survey of trust models in P2P environments is made. The newly-proposed models are analyzed, classified and compared.2. Distance-weighted Trust Model (DTM), a global trust model, is presented, which includes a mathematical description and a distributed implementation. Trust models in P2P environments have two main challenges: various malicious behavior of peers such as providing fake or misleading recommendation about other peers; decentralized and secure trust data management. Previous global trust models are based on the assumption that the peers with high trust value will give the honest recommendation. But this assumption may not hold in all cases. In DTM, each peer is assigned a unique global trust value, computed by aggregating distant-weighted recommendations of the peers who have interacted with it. Theoretical analyses and experiments show that the scheme is still robust under more general conditions where malicious peers cooperate in an attempt to deliberately subvert the system, converges more quickly and decreases the numbers of inauthentic files downloaded more effectively than previous schemes. Furthermore, a decentralized trust data management scheme is given and the anonymity requirements of peers for securing trust data are also considered.更多还原