Abstract

With the rapid growth of P2P network, the resource searching and resource delivery are two crucial problems required to be solved in a P2P system. At the resource searching stage, a large number of random resource requests form a search request queue at the node. At the resource delivery stage, if the nodes cannot process new requests at a certain time, the queue of resource requests may be congested at the node. The behavior of each user making a search request is a random phenomenon, so it is necessary to use the knowledge of queuing theory to allocate the users who request different resources and manage these request queues to provide services efficiently, which are based on the hybrid P2P network model in this paper. Firstly, a two-dimensional Markov chain is constructed, and the steady state distribution of system is determined by using the method of matrix geometric solution. Secondly, the expressions of performance indexes, such as the probability that the local Peer Node (PN), the remote PN and Virtual Content Server (VCS) provide the users, are obtained, and the influence of different system parameters on the performance indexes are analyzed using the results of numerical calculations. Finally, in order to avoid congestion of the request queue, the social benefit function is defined, and the optimal user arrival rate of system is obtained.

中文翻译：

---

P2P网络系统中基于请求队列的排队模型研究

摘要

随着P2P网络的快速发展，资源搜索和资源交付是P2P系统需要解决的两个关键问题。在资源搜索阶段，大量的随机资源请求在节点上形成一个搜索请求队列。在资源交付阶段，如果节点在某个时间无法处理新的请求，则资源请求队列可能会在该节点拥塞。每个用户发起搜索请求的行为是一种随机现象，因此需要利用排队论的知识来分配请求不同资源的用户并管理这些请求队列以高效地提供服务，这是基于混合P2P的本文中的网络模型。首先构造一个二维马尔可夫链，采用矩阵几何解法确定系统的稳态分布。其次，得到本地Peer Node(PN)、远端PN和Virtual Content Server(VCS)提供给用户的概率等性能指标的表达式，得到不同系统参数对性能指标的影响使用数值计算的结果进行分析。最后，为了避免请求队列的拥塞，定义了社会效益函数，得到系统的最优用户到达率。并利用数值计算结果分析了不同系统参数对性能指标的影响。最后，为了避免请求队列的拥塞，定义了社会效益函数，得到系统的最优用户到达率。并利用数值计算结果分析了不同系统参数对性能指标的影响。最后，为了避免请求队列的拥塞，定义了社会效益函数，得到系统的最优用户到达率。