流量矩阵分析的新方法研究
发布时间：2018-11-18 21:48
【摘要】：互联网技术作为21世纪发展最快的技术之一,已经广泛的应用于我们的生产生活当中,并且对社会的进步、经济的发展做出了巨大的贡献。然而,随着互联网技术进一步成熟,近年来也涌现出了大量的新型网络应用和服务,它们给人们带来方便娱乐的同时,也给网络运营商的管理维护带来了巨大的压力。与此同时,数量众多的异构网络的接入,更加使得互联网变得难以掌控。如何有效的监控和分析互联网络则显得尤为必要。 网络流量工程中的一个重要的参数就是流量矩阵,它对流量工程的重要性使得它广受研究人员的关注,并成为Internet的一个重要研究方向。流量矩阵的研究分为两个方面,流量矩阵的估计和流量矩阵的分析。本文将采用近年来新提出的一种分析方法来研究分析流量矩阵,并以此实现对流量矩阵异常的检测分析。本文的研究内容主要分为如下三个方面： 1)算子的选择。经过实验分析,不同的扩散小波算子将对小波系数矩阵产生微妙的变化,而这些变化将在一定程度上影响流量矩阵不同情况下的分析。所以本文的第一个工作将是设计实验,并分析对比两个常用的扩散小波算子,RandomWalk算子和I-L算子,然后选一个作为本文异常检测实验的扩散小波算子。文中设计了3个方向的对比实验来凸显两个算子各自的优劣。 2)异常检测。在完成扩散小波算子的对比实验后,本文将展开流量矩阵的异常检测实验。在异常检测实验中,本文将从异常检测算法设计和异常实验数据选择两方面展开,并给出最终的异常检测结果。 3)异常定位。在文章的最后,本文通过实验及统计,分析了扩散小波系数矩阵与原始流量矩阵之间存在的一些规律,通过这个规律可以由系数矩阵的异常变化来推测出原始流量矩阵中出现异常的节点的位置。作为对这个规律的应用,本文设计实验完成了流量矩阵的断路检测。 基于扩散小波的多尺度流量矩阵分析能够通过合适尺度的小波系数矩阵来解析原始流量矩阵信息。这样不但减少了分析的计算量,还能使分析变得更加准确有效。扩散小波算子的应用,使得流量矩阵的重要特征可以用小波系数矩阵来描述,两者之间存在的潜在联系对于网路工程中的应用都具备极大的研究价值。
[Abstract]:As one of the fastest developing technologies in the 21st century, Internet technology has been widely used in our production and life, and has made a great contribution to social progress and economic development. However, with the further maturity of Internet technology, a large number of new network applications and services have emerged in recent years, which bring people convenient entertainment, but also bring great pressure to the management and maintenance of network operators. At the same time, a large number of heterogeneous networks access, making the Internet more difficult to control. How to effectively monitor and analyze the Internet is particularly necessary. Traffic matrix is an important parameter in network traffic engineering. Its importance to traffic engineering makes it widely concerned by researchers and becomes an important research direction of Internet. The research of the flow matrix is divided into two aspects: the estimation of the flow matrix and the analysis of the flow matrix. In this paper, a new analysis method proposed in recent years is used to study and analyze the flow matrix and to detect and analyze the anomaly of the flow matrix. The main contents of this paper are as follows: 1) selection of operators. Through experimental analysis, different diffusive wavelet operators will produce subtle changes to the wavelet coefficient matrix, and these changes will influence the analysis of the flow matrix under different conditions to some extent. Therefore, the first work of this paper will be to design experiments and compare two diffusive wavelet operators, RandomWalk operators and I-L operators, and then select one as the diffusive wavelet operator of anomaly detection experiment in this paper. A comparative experiment in three directions is designed to highlight the advantages and disadvantages of the two operators. 2) abnormal detection. After the contrast experiment of diffusive wavelet operator is completed, the anomaly detection experiment of flow matrix will be carried out in this paper. In the experiment of anomaly detection, the algorithm design of anomaly detection and the data selection of anomaly experiment are discussed in this paper, and the final results of anomaly detection are given. 3) abnormal location. At the end of the paper, some laws between the diffusion wavelet coefficient matrix and the original flow matrix are analyzed through experiments and statistics. According to this rule, the abnormal position of the node in the original flow matrix can be deduced from the abnormal change of the coefficient matrix. As an application of this rule, this paper designs experiments to complete the open circuit detection of the flow matrix. Multi-scale traffic matrix analysis based on diffusive wavelet can analyze the information of original flow matrix by wavelet coefficient matrix of appropriate scale. This not only reduces the calculation of the analysis, but also makes the analysis more accurate and effective. With the application of diffusive wavelet operator, the important characteristics of traffic matrix can be described by wavelet coefficient matrix. The potential relationship between them has great value for the application of network engineering.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.06
本文编号：2341360
[Abstract]:As one of the fastest developing technologies in the 21st century, Internet technology has been widely used in our production and life, and has made a great contribution to social progress and economic development. However, with the further maturity of Internet technology, a large number of new network applications and services have emerged in recent years, which bring people convenient entertainment, but also bring great pressure to the management and maintenance of network operators. At the same time, a large number of heterogeneous networks access, making the Internet more difficult to control. How to effectively monitor and analyze the Internet is particularly necessary. Traffic matrix is an important parameter in network traffic engineering. Its importance to traffic engineering makes it widely concerned by researchers and becomes an important research direction of Internet. The research of the flow matrix is divided into two aspects: the estimation of the flow matrix and the analysis of the flow matrix. In this paper, a new analysis method proposed in recent years is used to study and analyze the flow matrix and to detect and analyze the anomaly of the flow matrix. The main contents of this paper are as follows: 1) selection of operators. Through experimental analysis, different diffusive wavelet operators will produce subtle changes to the wavelet coefficient matrix, and these changes will influence the analysis of the flow matrix under different conditions to some extent. Therefore, the first work of this paper will be to design experiments and compare two diffusive wavelet operators, RandomWalk operators and I-L operators, and then select one as the diffusive wavelet operator of anomaly detection experiment in this paper. A comparative experiment in three directions is designed to highlight the advantages and disadvantages of the two operators. 2) abnormal detection. After the contrast experiment of diffusive wavelet operator is completed, the anomaly detection experiment of flow matrix will be carried out in this paper. In the experiment of anomaly detection, the algorithm design of anomaly detection and the data selection of anomaly experiment are discussed in this paper, and the final results of anomaly detection are given. 3) abnormal location. At the end of the paper, some laws between the diffusion wavelet coefficient matrix and the original flow matrix are analyzed through experiments and statistics. According to this rule, the abnormal position of the node in the original flow matrix can be deduced from the abnormal change of the coefficient matrix. As an application of this rule, this paper designs experiments to complete the open circuit detection of the flow matrix. Multi-scale traffic matrix analysis based on diffusive wavelet can analyze the information of original flow matrix by wavelet coefficient matrix of appropriate scale. This not only reduces the calculation of the analysis, but also makes the analysis more accurate and effective. With the application of diffusive wavelet operator, the important characteristics of traffic matrix can be described by wavelet coefficient matrix. The potential relationship between them has great value for the application of network engineering.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.06
【参考文献】
相关期刊论文 前10条
1 蒋定德;胡光岷;倪海转;;IP骨干网络流量矩阵估计算法研究[J];电子科技大学学报;2010年03期
2 段丽英;符蕴芳;李建波;;网络异常入侵检测研究[J];福建电脑;2006年08期
3 郑淋;叶猛;;基于多尺度分析和决策树的P2P流量检测模型[J];电视技术;2013年01期
4 王意志;王呈炎;;网络测量方法及其应用[J];成都大学学报(自然科学版);2012年04期
5 蒋定德;胡光岷;;流量矩阵估计研究综述[J];计算机科学;2008年04期
6 张科;谢佳;胡光岷;邓正虹;;基于广义线性反演的流量矩阵估计算法[J];计算机应用;2008年03期
7 刘兰;李之棠;李家春;谭晓玲;;小波及网络异常行为分析[J];计算机应用研究;2007年04期
8 狄剑光;陈光英;孙东红;;网络异常检测[J];中国教育网络;2006年05期
9 季凯;;重力模型标定方法及分析[J];山西建筑;2012年11期
10 魏多;吕光宏;;基于蚁群算法的IP网络流量矩阵估计[J];计算机应用;2013年01期
，本文编号：2341360
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2341360.html
