P2P网络中一种基于进化博弈的资源配置模型  

Evolutionary Game Theoretical Resource Deployment Model for P2P Networks

作　　者：周经亚[1] 宋爱波[1] 罗军舟[1] 

机构地区：[1]东南大学计算机科学与工程学院,江苏南京211189

出　　处：《软件学报》2013年第3期526-539,共14页Journal of Software

基　　金：国家重点基础研究发展计划(973)(2010CB328104);国家自然科学基金(61070161;61202449;61272054;61003257);国家科技支撑计划(2010BAI88B03;2011BAK21B02);高等学校博士点学科专项科研基金(20110092130002);国家科技重大专项科研基金(2010ZX01044-001-001);江苏省自然科学基金(BK2008030);江苏省产学研前瞻性联合研究项目(BY2012202);江苏省科技成果转化专项资金项目(BA2012036);江苏省网络与信息安全重点实验室资助项目(BM2003201);教育部计算机网络与信息集成重点实验室(东南大学)资助项目(93K-9);上海市可扩展计算与系统重点实验室(上海交通大学)资助项目(2010DS680095);浙江师范大学计算机软件与理论省级重中之重学科开放基金(ZSDZZZZXK34);中国教育科研网格ChinaGrid资助项目

摘　　要：合理的资源配置能够有效地改进非结构化P2P网络的查询性能,提高资源副本的可获得性.当前,资源配置研究多集中在各种类型资源副本的定量分析和分布式配置策略上,节点独立地选择资源副本进行配置,并未考虑节点间配置行为的交互作用.P2P网络中节点只维护若干与邻居节点的连接,掌握局部信息,因而在交互过程中可将节点视为有限理性节点.在分析查询性能与节点资源配置行为之间关系的基础上,构造查询性能相关的节点收益函数,将资源配置问题模型化为一种进化博弈,通过对进化过程的描述能够有效分析节点在资源配置过程中的交互关系以及可获得的查询性能.仿真实验结果表明,资源配置进化模型可获得更高的查询成功率和近似最优的平均查询跳数,且保持相对较低的冗余度.Resource deployment is an effective means to improve search performance and can also be used to enhance the availability of resource replicas in unstructured P2P networks. Most of the current studies focus on the quantitative analysis of various types of resource replicas and distributed deployment strategies. During the resource deployment process each node selects resource replica exclusively for deployment; however, the process lacks a consideration for deployment behavior interactions among participating nodes. In a P2P network, each node keeps in touch with several other neighbors and are aware of local information, so each node can be assumed to be bounded rational. This paper designs the performance-related payoff function through analyzing the relation between search performance and resource deployment behaviors of nodes, and then models the resource deployment as an evolutionary game. In terms of the description of game evolution, the study can effectively analyze the interactions among nodes and the expected search performance. The simulation results indicate that the proposed resource deployment evolutionary model achieves higher success rate and approximate optimal average hop counts while maintaining a relatively low redundancy.

关 键 词：对等网络 广义随机图 进化博弈 资源配置 泛洪 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]