【作者】 张微寒；

【导师】 张学杰；

【作者基本信息】 云南大学 ， 计算机应用技术， 2011， 硕士

【摘要】 由于缺乏有效的激励手段,传统的P2P (Peer-to-Peer)网络普遍存在着搭便车现象以及“公共物品的悲剧”等问题,严重影响了网络的健壮性及可用性。因此,如何构建和设计合理的激励机制成为了当前P2P网络研究中的一个热点问题。在P2P文件共享系统中,基于博弈论的激励机制有效地抑制了搭便车行为,但该机制容易引起whitewashing现象；而在P2P流媒体系统中,当前的激励机制没有综合考虑节点的贡献以及对资源的需求程度,因而出现了系统资源的不公平调度等问题。因此,研究P2P网络中的激励机制,具有重要的学术意义和应用价值。针对上述问题,本文以文件共享系统和流媒体系统为研究对象,以经济学中的博弈理论为指导,对P2P网络中的激励机制进行了分析和研究。一方面,针对P2P文件共享系统中的whitewashing问题,引入完全信息、博弈中的均衡选择博弈模型对P2P网络进行建模,该博弈模型的优势在于它不依赖于统一的行动信号,使其过程可更有效地模拟P2P网络中节点的随机请求；此外,均衡选择博弈具有多个均衡解,其最大最小策略在系统保持均衡的状态下能有效的解决新加入节点的下载请求问题,弥补了传统激励模型的不足。模拟实验表明,所提激励机制可有效促进系统资源的公平分发,同时提高了节点对网络的贡献度,有效地约束了自私节点的行为,保障了节点和系统的效用。另一方面,针对流媒体系统中文件播放实时性、节点传输有序性等特点,本文以不完全信息博弈的相关理论为指导对P2P流媒体系统中的激励问题展开研究。特别地,引入一级密封价格拍卖机制对系统的激励模型进行分析和建模,使系统中的服务节点根据请求节点的贡献度以及所申请资源的紧急程度,响应其资源请求。最后,我们对CoolStreaming系统进行了改进和优化,加入了激励模块。实验结果表明,所提激励模型可有效提升流媒体系统中资源调度的效率,保障了节点的服务质量。更多还原

【Abstract】 As lacking of effective incentives, there exist some ubiquitous problems such as free-riding and’the tragedy of the commons’phenomenon in the traditional P2P (Peer-to-Peer) network which seriously affect the network’s robustness and availability. Therefore, how to design and construct efficient incentive mechanism has became one of the hotspot issues of P2P studies. The game theory has been proved to be effectively restrained the free-riding phenomenon in the file sharing system. However, the incentives based on game theory can lead to whitewashing problem. Also in P2P streaming media system, the current incentive mechanism is insufficient in satisfying the requirement for resources, and the node’s contribution is neglected. As a consequence, there will be some problems such as unfair scheduling of resources etc. Thus, it has a great academic significance and application value that doing research on the incentive mechanism in P2P networks.Based on the above issues, this thesis analysis the incentive mechanism in P2P network using game theory of economics as guide, and studies of file sharing system and streaming media system as background. On one hand, the equilibrium selection game model of the complete information game is introduced to solve the whitewashing problem. The advantage of this model is that it does not dependent on unified action signals, which makes its process more efficient for simulating the random request of the network node. Moreover, this system has more than one strategies, which the maxmin strategy compensates the deficiency of the traditional incentive system by resolving the new joined node’s download request while keeping the system’s equilibrium. Simulation experiments shows that this incentive mechanism is not only able to promote the fair distribution of the system resources, it is also increased the contribution of the nodes by restrain the selfish nodes and ensures the system utility.On the other hand, according to the real-time request of playback time and the sequential transmission of media file, this thesis studies on the incentive mechanism in streaming media system with incomplete information game as guide. Especially, the first-price sealed big auction is introduced to modeling the incentive mechanism which makes the resources distribution according to the level of the node’s contribution and the urgency of the resource request. Finally, we has improved and optimized the CoolStreaming system, and added the incentive module. The result of the experiment indicates that the proposed incentive mechanism is able to enhance the scheduling efficiency, and ensure the QoS of the nodes.更多还原