基于实测复杂网络模型的弹性研究
发布时间：2018-09-07 19:54
【摘要】：尽管网络的发展迅速,每天都会有成千上万的网络服务器开启或者是关闭,网络在运行中还经常受到干扰或破坏,从而导致网络本身性能降低甚至是功能瘫痪。人们对于网络的依赖程度日益剧增,但网络事故对人们产生的打击亦是随着依赖的增加而增加。从北美大停电到20世纪爆发的金融危机,无不透漏着我们对网络的依赖和网络瘫痪对我们照成的巨大损失。对于互联网络而言,弹性结构是当下评价网络体统的新想法。在之前很多研究中,很少有学者对实测复杂网络的弹性进行深入的研究,对复杂网络的研究也仅仅局限于网络自身恢复能力的理论验证,但这些问题对于实测复杂网络如何抵御攻击并减少故障有着重要意义,因此本文研究重点在于对复杂网络结构模型及互联网弹性进行深入探讨。本文主要做了以下三个方面的工作:1.对复杂网络的理论分析,首先通过对复杂网络模型进行深入研究,分别对规则网络、随机网络、小世界和无标度网络进行构建并分析其特有性质。随后对网络统计特性指标度分布、平均路径长度和网络聚类系数进行详细介绍,并通过实验模拟对不同网络属性进行分析。2.详细介绍了网络实际测量方法和测量指标,对现有国内外的资料进行分析,选取性价比最高的测量方法对互联网进行测量并为特性分析提供数据。通过引入网络弹性定义,对测量数据进行实验。实验结果表明:ER随机网络对于恶意攻击弹性要好于其他网络;BA无标度网络弹性恢复较好;规则网络恢复弹性表现最差;WS小世界则特性并不明显,其弹性介于ER随机网络和规则网络之间。而在互联网络中,弹性恢复主要受恢复措施影响较大,弹性连接并不能够使网络完全恢复。3.考虑网络静态特性,从差异性角度分析网络结构,并引入“熵”指标对规则网络、随机网络、无标度网络和小世界进行理论分析和仿真实验。通过实验数据得出熵在复杂网络中更能反映出其结构特征。
[Abstract]:Despite the rapid development of the network, thousands of network servers are opened or shut down every day, and the network is often disturbed or destroyed in operation, which results in the performance of the network itself being degraded or even paralyzed. The degree of people's dependence on network is increasing rapidly, but the attack of network accident is also increasing with the increase of dependence. From the power outages in North America to the financial crisis that broke out in the 20th century, all of us have been exposed to the enormous losses caused by our dependence on the network and the collapse of the network. For the Internet, flexible structure is a new idea to evaluate the network system. In many previous studies, few scholars have carried out in-depth research on the elasticity of measured complex networks, and the research on complex networks is limited to the theoretical verification of the resilience of the networks themselves. However, these problems are of great significance to how to resist attacks and reduce faults in real complex networks. Therefore, the focus of this paper is to discuss the complex network structure model and Internet elasticity in depth. This paper mainly does the following three aspects of work: 1. Based on the theoretical analysis of complex networks, this paper studies the model of complex networks, constructs regular networks, random networks, small world networks and scale-free networks, and analyzes their special properties. Then, the distribution of network statistical characteristics, average path length and network clustering coefficient are introduced in detail, and the different network attributes are analyzed by experimental simulation. This paper introduces the network actual measurement method and measurement index in detail, analyzes the existing data at home and abroad, selects the best measurement method to measure the Internet and provides the data for the characteristic analysis. By introducing the definition of network elasticity, the measurement data are tested. The experimental results show that the resilience of the 10: ER random network to malicious attack is better than that of the other networks, and the resilience of the rule network is the worst, but the performance of the rule network is not obvious in the small world of WS. Its elasticity is between ER random network and regular network. In the Internet, the elastic recovery is mainly affected by the restoration measures, and the elastic connection can not make the network recover completely. 3. Considering the static characteristics of the network, the network structure is analyzed from the point of view of difference, and the "entropy" index is introduced to carry out theoretical analysis and simulation experiments on regular network, random network, scale-free network and small world. The experimental data show that entropy can better reflect the structural characteristics of complex networks.
【学位授予单位】：沈阳理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5
本文编号：2229260
[Abstract]:Despite the rapid development of the network, thousands of network servers are opened or shut down every day, and the network is often disturbed or destroyed in operation, which results in the performance of the network itself being degraded or even paralyzed. The degree of people's dependence on network is increasing rapidly, but the attack of network accident is also increasing with the increase of dependence. From the power outages in North America to the financial crisis that broke out in the 20th century, all of us have been exposed to the enormous losses caused by our dependence on the network and the collapse of the network. For the Internet, flexible structure is a new idea to evaluate the network system. In many previous studies, few scholars have carried out in-depth research on the elasticity of measured complex networks, and the research on complex networks is limited to the theoretical verification of the resilience of the networks themselves. However, these problems are of great significance to how to resist attacks and reduce faults in real complex networks. Therefore, the focus of this paper is to discuss the complex network structure model and Internet elasticity in depth. This paper mainly does the following three aspects of work: 1. Based on the theoretical analysis of complex networks, this paper studies the model of complex networks, constructs regular networks, random networks, small world networks and scale-free networks, and analyzes their special properties. Then, the distribution of network statistical characteristics, average path length and network clustering coefficient are introduced in detail, and the different network attributes are analyzed by experimental simulation. This paper introduces the network actual measurement method and measurement index in detail, analyzes the existing data at home and abroad, selects the best measurement method to measure the Internet and provides the data for the characteristic analysis. By introducing the definition of network elasticity, the measurement data are tested. The experimental results show that the resilience of the 10: ER random network to malicious attack is better than that of the other networks, and the resilience of the rule network is the worst, but the performance of the rule network is not obvious in the small world of WS. Its elasticity is between ER random network and regular network. In the Internet, the elastic recovery is mainly affected by the restoration measures, and the elastic connection can not make the network recover completely. 3. Considering the static characteristics of the network, the network structure is analyzed from the point of view of difference, and the "entropy" index is introduced to carry out theoretical analysis and simulation experiments on regular network, random network, scale-free network and small world. The experimental data show that entropy can better reflect the structural characteristics of complex networks.
【学位授予单位】：沈阳理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5
【参考文献】
相关期刊论文 前4条
1 蔡萌;杜海峰;任义科;费尔德曼;;一种基于点和边差异性的网络结构熵[J];物理学报;2011年11期
2 王延;郑志刚;;无标度网络上的传播动力学[J];物理学报;2009年07期
3 张宇,张宏莉,方滨兴;Internet拓扑建模综述[J];软件学报;2004年08期
4 谭跃进,吴俊;网络结构熵及其在非标度网络中的应用[J];系统工程理论与实践;2004年06期
，本文编号：2229260
本文链接：https://www.wllwen.com/kejilunwen/yysx/2229260.html
