基于随机抽样过程的P2P集群规模估算方法  

Estimation of P2P Swarm Size Based on Random Sampling Process

作　　者：王潇斌[1] 李程[1] 石碧[1] 杨哲[1] 

机构地区：[1]苏州大学计算机科学与技术学院,江苏苏州215006

出　　处：《南京师大学报（自然科学版）》2014年第1期76-80,98,共6页Journal of Nanjing Normal University(Natural Science Edition)

基　　金：国家自然科学基金(61070170);江苏省高校自然科学基金(11KJB520017);苏州市科技计划项目(SYG201238;SZS0805)

摘　　要：对P2P系统的主动测量,可了解其现状和变化趋势,为系统建模和仿真提供可靠的测量依据.现有的赠券收集者模型,过度依赖于服务器返回的先验知识,导致测量结果不能反映集群的真实规模.基于随机抽样过程,本文提出了一种P2P集群规模主动估算方法.根据测量过程中不同时刻获取的节点总数x及不重复节点数u,得到集群规模的估计值.根据理论分析的结果,分别给出了3种不同的实验停止条件.实验结果表明,对于小于105的集群,本文的估计方法误差不超过5%.Active measurement on peer-to-peer system is the best way to understand the current situation and its evolution trends. It also provides the real measurement basis to support the research on system simulation and optimization. The coupon collectors model is too dependent on the prior knowledge returned by server,so the measurement results do not reflect the true scale of the P2P swarm. Based on random sampling process,this paper presents a P2P active swarm size estimation method. Based on the total number of nodes x and the unique nodes u,it can estimate the P2P swarm size. Ac-cording to the theoretical analysis of the results, we give three different experiments stop condition. The experimental results show that for less than 105 clusters,this estimation method error does not exceed 5%.

关 键 词：P2P集群 主动测量 随机抽样过程 集群规模估计 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]