基于在线概率的P2P文件共享系统的动力学建模

发布时间：2018-03-24 12:42

本文选题：P2P　切入点：文件共享系统　出处：《中国科学技术大学》2017年硕士论文

【摘要】：随着互联网的不断普及和网络技术的迅猛发展,P2P(peerto peer)文件共享技术得到了越来越广泛的应用。据统计,P2P文件共享已成为承载大部分网络流量的主流互联网应用。因此,对P2P文件共享系统进行建模研究具有重要的意义。在P2P文件共享系统中,加入系统的所有用户节点构成了一个复杂网络。整个文件共享网络无中心控制机制,按照一种自组织的方式运行。由于网络中用户行为的随机性以及其它一些随机因素的影响,用户节点加入和退出网络也是随机的。节点的加入和退出会引起系统拓扑结构的变化。为了更加精确地刻画节点行为的随机性,更加准确地描述系统的演化过程,本文在系统建模中引入了节点在线概率的概念,建立了基于在线概率的动力学模型。本文所做的工作主要集中在以下几个方面:第一,研究了 P2P文件共享系统中的用户行为问题。首先,本文运用泊松分布分析了系统中节点的上下线行为,并且给出了节点加入系统的上线概率表达式。然后,我们对节点加入和退出系统的行为进行分析,得出了节点在线概率近似服从正态分布的结论。第二,研究了基于在线概率的动力学建模问题。本文从借贷关系的角度出发,对P2P文件共享系统的运行过程进行了分析,定义了系统模型中的有关变量。然后,分析了影响系统演化的主要因素,包括用户行为、文件属性、硬件条件和软件算法,随后,通过引入在线概率以及对系统演化过程的分析,建立了基于在线概率的P2P文件共享系统的动力学模型。第三,对P2P文件共享系统中的相关算法进行改进,提出了基于在线概率的节点选择算法、带宽分配算法和节点阻塞算法。进一步,我们根据算法的具体表达式得到了系统具体的基于在线概率的动力学方程。第四,通过仿真实验,验证了基于在线概率的动力学模型的正确性。本文首先设计了两个仿真实例,分别分析了当系统节点上传带宽远大于文件请求速率和节点上传带宽远小于文件请求速率时的动力学模型中的节点状态演化曲线,进而对动力学模型的正确性进行了验证。然后通过对采用等概率算法和在线概率算法的动力学模型分别进行仿真分析,验证了基于在线概率动力学模型的正确性。最后对当在线概率服从不同正态分布时的系统动力学模型进行仿真,分析不同在线概率对系统演化过程的影响,进而对在线概率越大,节点的状态演化曲线的抖动频次越小,完成文件传输所需的时间越短这一结论的正确性进行了验证。
[Abstract]:With the continuous popularization of the Internet and the rapid development of network technology, the P2P file sharing technology has been more and more widely used. According to statistics, P2P file sharing has become the mainstream Internet application carrying most network traffic. It is of great significance to model P2P file sharing system. In P2P file sharing system, all users who join the system form a complex network. The file sharing network has no central control mechanism. Run in a self-organized manner. Because of the randomness of user behavior in the network and other random factors, In order to describe the randomness of node behavior more accurately and accurately describe the evolution process of the system, the user nodes join and exit the network is also random. In this paper, the concept of node online probability is introduced in system modeling, and a dynamic model based on online probability is established. The work in this paper is mainly focused on the following aspects: first, In this paper, the problem of user behavior in P2P file sharing system is studied. Firstly, Poisson distribution is used to analyze the upper and lower line behavior of nodes in the system. We analyze the behavior of node join and exit system, and get the conclusion that the node online probability approximate from normal distribution. Secondly, the dynamic modeling problem based on online probability is studied. This paper analyzes the running process of P2P file sharing system, defines the relevant variables in the system model, and then analyzes the main factors that affect the evolution of the system, including user behavior, file attributes, hardware conditions and software algorithms. By introducing online probability and analyzing the evolution process of P2P file sharing system, the dynamic model of P2P file sharing system based on online probability is established. Thirdly, the related algorithms in P2P file sharing system are improved. The node selection algorithm, bandwidth allocation algorithm and node blocking algorithm based on online probability are proposed. Furthermore, according to the specific expression of the algorithm, we obtain the specific dynamic equation of the system based on online probability. The correctness of the dynamic model based on online probability is verified by simulation experiments. Two simulation examples are designed in this paper. The node state evolution curves in the dynamic model are analyzed when the system node upload bandwidth is much larger than the file request rate and the node upload bandwidth is much lower than the file request rate. Then, the correctness of the dynamic model is verified. Then, the dynamic model with equal probability algorithm and online probability algorithm is simulated and analyzed, respectively. The correctness of the online probabilistic dynamics model is verified. Finally, the system dynamics model with different normal distribution is simulated, and the influence of different online probabilities on the evolution of the system is analyzed. Furthermore, the correctness of the conclusion that the higher the online probability, the smaller the jitter frequency of the state evolution curve and the shorter the time required to complete the file transfer are verified.
【学位授予单位】：中国科学技术大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.0

【相似文献】