混沌理论用于预测网络游戏流量的研究

发布时间：2018-07-18 13:08

【摘要】：近些年由于计算机硬件技术以及网络通信技术的飞速发展,使得电子游戏产业越来越成熟,其中发展最快的是网络游戏行业,巨大的网络市场使得网络游戏的发展受到了社会各界的关注,尤其是越来越多的IT企业以及游戏开发商开始投入到网络游戏行业,同时网络游戏的玩家也越来越多,随之带来的是网络游戏的服务器在性能和负载上遇到了很大的压力。如果你想要给网络游戏玩家提供最佳的服务并且降低游戏运行和维护成本的关键就是控制好服务器的架构以及负载均衡。本文选择了以网络游戏流量的预测为研究对象,研究了网络流量的分形特性,本文的主要研究内容包括:首先对采集的网络流量数据进行分析,对流量数据进行相空间重构,计算关联维,从而分析流量时间序列的分形特性,实现时间延迟的确定,得到自相关函数,初步确定网络流量的混沌特性。然后进行相空间重构,确定最小嵌入维和网络流量时序的宏观参数---关联维,从而实现网络流量行为特性的定量描述。再次,验证了网络流量信息系统的动力学行为是混沌的以及它的吸引子表现为分形结构。基于网络流量行为具有分形特性,基于相空间重构理论,建立了一个混沌模型,用来进行网络流量的预测;接着对建立的模型进行了仿真和验证;验证结果表明,模型的预测值与真实值的相对误差在0.02到0.08之间;之后利用模型对作者所在某游戏公司的某游戏服务器群组进行了监测,预测精度在0.02到0.11之间。文中所研究的网络流量的分形特性及相关预测模型的设计与实践分析,对了解和解决网络游戏流量,做好服务器组的负载均衡及配置具有较好的实用参考价值。
[Abstract]:In recent years, due to the rapid development of computer hardware technology and network communication technology, the electronic game industry has become more and more mature, among which the fastest growing industry is the online game industry. The huge network market makes the development of the network game receive the attention of all circles of the society, especially more and more IT enterprises and game developers begin to invest in the network game industry, at the same time, the players of the network game are also more and more. What brings along is the network game server in the performance and the load has encountered the very big pressure. If you want to provide the best service to online gamers and reduce the cost of running and maintaining the game, the key is to control the server architecture and load balancing. In this paper, the prediction of network game traffic is selected as the research object, and the fractal characteristics of network traffic are studied. The main contents of this paper are as follows: firstly, the collected network traffic data are analyzed, and the phase space of traffic data is reconstructed. The correlation dimension is calculated, and the fractal characteristics of traffic time series are analyzed, the time delay is determined, the autocorrelation function is obtained, and the chaotic characteristics of network traffic are preliminarily determined. Then the phase space reconstruction is carried out to determine the macroscopical parameter-correlation dimension of the minimum embedding and network traffic timing, so that the quantitative description of network traffic behavior can be realized. Thirdly, it is verified that the dynamical behavior of network traffic information system is chaotic and its attractor is fractal structure. Based on the fractal characteristics of network traffic behavior, a chaotic model is established based on phase space reconstruction theory, which is used to predict network traffic. The relative error between the predicted value and the real value of the model is between 0.02 and 0.08, and then the model is used to monitor the game server group in a game company, and the prediction accuracy is between 0.02 and 0.11. The fractal characteristics of network traffic and the design and practical analysis of the related prediction model are helpful to understand and solve the network game traffic and to balance and configure the load of server group.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.06;O415.5

【共引文献】