无结构对等网络激励机制研究

【作者】 康江；

【导师】 房鼎益；

【作者基本信息】 西北大学 ， 计算机软件与理论， 2010， 硕士

【摘要】 当今因特网应用中出现了越来越多的基于对等网络协议开发的应用软件,其网络结构一般分为结构化和无结构化两大类,其中无结构对等网络因其本身具有自治性、无组织、去中心化等特点而倍受关注。本文的研究对象是无结构对等网络系统中广泛存在的搭便车问题。在充分研究近年来国内外已有的多种激励机制模型后,我们发现现有的P2P (Peer-to-Peer)应用软件的运营模式与已经提出的众多策略下的激励机制之间存在不一致性。激励机制主要关注的是如何通过奖励协作节点、抑制搭便车节点来维护整个网络系统的公平公正性,而P2P网络运营商的主要收入是来自在网络社区内广告的投放量,因此为了吸引更多的用户加入P2P网络并长时间留在网络社区中,运营商往往对搭便车节点采取了忽视和容忍的态度。针对上述问题,本文借鉴了分布式测量和博弈论中的纳什均衡理论,提出DAMR(Distributed algorithm Anti-free-rider based on Message Routing)算法,该算法分布式地检测和抑制搭便车节点。在此算法的基础上,考虑节点自身的网络负载,设计了一个基于DAMR策略的激励机制。为了更好地分析本文提出的激励机制的特点、性能、对P2P网络系统的影响及其是否符合P2P网络运营商的商业期望,通过数学化建模的方法,提出一个无结构P2P网络系统激励机制的数学评估模型,并利用该模型分析论证了基于DAMR策略的激励机制能够使P2P网络系统处于一种良性的均衡状态。最后通过扩展GnutellaSim软件,将该激励机制添加到以NS-2为平台的仿真系统中,通过实验和数据分析,验证了基于DAMR策略的激励机制的正确性和有效性。利用该激励机制,在符合P2P网络运营商利益的前提下,可以提高网络的利用率,促进Peer节点参与协作和享用服务,使P2P网络社区更具有吸引力。本文的研究对促进P2P网络系统的良性发展具有一定的启示意义。更多还原

【Abstract】 More and more applications based on Peer-to-Peer protocol spring up on the Internet, their network structures are two main categories:structural networks and unstructured networks, people put more attention to the unstructured Peer-to-Peer (P2P) networks because they are autonomous, unstructured and decentralized. The research object of the paper is the free-riding problem which widely exists in the unstructured P2P networks. On the basis of study on the classical incentive mechanisms proposed in recent years, we find that there is a conflict between the operating mode on which most P2P applications are currently running and the aim of the proposed incentive mechanisms. The proposed incentive mechanisms focus mainly on how to reward cooperative nodes punish the free riders so as to make the entire network be justice and equitable. However, main income of P2P network operator comes from advertisements in the network community, so in order to attract more users to join the network and stay in network for a long time, operators tend to take the attitude of neglecting and tolerance to free-riders.Aiming at the problem, through drawing on the distributed measurement and the Nash equilibrium in game theory, a DAMR(Distributed algorithm Anti-free-rider based on Message Routing) algorithm is proposed in the paper。This algorithm detects and discourage the free-riders distributively。On the basis of the DAMR algorithm, an incentive mechanism is designed by taking the node’s network load into account. In order to analyze the characteristics and performance of the incentive mechanism proposed in this paper better, and to analyze the influence of the incentive mechanism to P2P network and operators, an evaluation model is proposed. Making use of this mathematical model, we evaluate the DAMR incentive mechanism and draw the conclusion that the incentive mechanism can make P2P network system to reach and keep a positive equilibrium. At the end of the paper, through the expansion of GnutellaSim software to add the incentive mechanism to NS-2, a platform for simulation is built up. Through making many times of simulation experiments, a series of data are got. By analysis of the data, the correctness and effectiveness of the DAMR-based incentive mechanism are verified. So the conclusion is that the DAMR-based incentive mechanism can make more efficient use of network, promote the nodes in P2P network to collaborate and enjoy services, make P2P networks more attractive in accordance with the interests of P2P network operators. The research in the paper may have some inspiration to the sound development of P2P networks.更多还原