复杂网络抗攻击能力优化问题的研究
发布时间：2018-09-06 13:04
【摘要】：随着互联网技术的蓬勃发展及其日益广泛地被应用于人们的日常生活,网络攻击事件层出不穷,如银行客户信息泄漏、联想平板低价遭哄抢、根服务器故等,因此加强网络保护、构建高鲁棒性的网络系统至关重要。复杂网络作为有力工具能够很好地刻画真实世界中复杂系统的结构特征和典型动力学行为;复杂网络的攻击和保护问题成为相关领域的一个研究热点。本文在复杂网络大背景下,综合应用数理统计、控制、计算机仿真等多学科领域知识,围绕“如何提升网络的抗攻击能力”、“网络抗攻击能力与同步能力的关系”、“聚簇系数对相依网络结构脆弱性的影响”这三个问题展开了深入研究,并取得了如下创新性研究成果:(1)基于模拟退火算法的网络抗攻击能力优化研究基于网络抗攻击能力优化问题,给出了调整网络结构的保度边交叉重连方法,建立了优化模型,将网络抗攻击能力优化问题转化为单目标组合优化问题,紧接着在求解单目标组合优化问题中引入了模拟退火算法,以求得全局范围内的最优解,最后通过数值仿真实验验证了优化算法的有效性。同时,还记录分析了结构优化过程中重要的结构特征参数的变化规律。(2)onion-like结构的动态无标度网络同步能力分析首先对网络的抗攻击能力进行了优化,观察发现,优化后的网络拓扑结构呈onion-like状,即度大的节点组成核心且从内层到外层节点度值依次递减;然后研究了具有onion-like结构的动态无标度网络同步稳定性。数值仿真结果表明随着网络抗攻击能力的提升,其同步能力会减弱。同时,分析了onion-like结构网络的两个典型结构参数,平均路径长度和度相配系数,从理论角度解释了抗攻击能力提升同时同步能力减弱的原因。(3)可调聚簇系数的相依网络结构脆弱性分析聚簇系数是影响网络性能的重要参数之一,提出了聚簇系数可调的相依无标度网络模型并采用了一种更为现实的攻击策略---针对度大节点的蓄意攻击。综合考虑了聚簇性和相依偏好两方面的因素,数值仿真结果表明两个网络之间相依边的存在使得网络抗攻击能力变弱,且高聚簇性会使得相依无标度网络系统更加脆弱。进一步来说,对全相依网络而言,异配型网络抗随机攻击能力最弱,但是随机型和同配型网络基本没有差别。此外,仿真结果表明相依度的增加会加剧整个系统的脆弱性。
[Abstract]:With the rapid development of Internet technology and its increasingly widespread application in people's daily life, network attacks emerge in endlessly, such as bank customer information leakage, Lenovo flat price being robbed, root server, and so on. Therefore, it is very important to strengthen the network protection and build a network system with high robustness. As a powerful tool, complex networks can well describe the structural characteristics and typical dynamic behaviors of complex systems in the real world, and the attack and protection of complex networks have become a research hotspot in related fields. Under the background of complex network, this paper synthetically applies the knowledge of mathematical statistics, control, computer simulation and so on, and revolves around "how to improve the anti-attack ability of the network", "the relationship between the network anti-attack ability and the synchronization ability", and so on. The influence of clustering coefficient on the vulnerability of dependent network structure is studied. The main achievements are as follows: (1) the optimization of network anti-attack ability based on simulated annealing algorithm; based on the optimization of network anti-attack ability, a method of preserving the network structure is presented, and the optimization model is established. The optimization problem of network anti-attack ability is transformed into a single-objective combinatorial optimization problem. Then the simulated annealing algorithm is introduced to solve the single-objective combinatorial optimization problem in order to obtain the global optimal solution. Finally, the effectiveness of the optimization algorithm is verified by numerical simulation. At the same time, the variation law of important structural characteristic parameters in the process of structural optimization is also recorded and analyzed. (2) the synchronization ability of dynamic scale-free network with onion-like structure is analyzed. Firstly, the anti-attack ability of the network is optimized and observed. The optimized network topology is onion-like, that is, the nodes with large degree form the core and decrease the degree from the inner layer to the outer layer. Then, the synchronization stability of the dynamic scale-free network with onion-like structure is studied. The numerical simulation results show that the synchronization ability of the network becomes weaker with the improvement of the network anti-attack ability. At the same time, two typical structural parameters, the average path length and the degree matching coefficient, of the onion-like network are analyzed. This paper theoretically explains the reason why the ability to resist attack is improved and the synchronization ability is weakened. (3) the dependent network structure vulnerability analysis of adjustable clustering coefficient is one of the important parameters that affect the network performance. The dependent scale-free network model with adjustable clustering coefficient is proposed and a more realistic attack strategy is proposed. Considering the factors of clustering and dependency preference, the numerical simulation results show that the existence of dependent edges between the two networks weakens the anti-attack ability of the network, and the high clustering makes the dependent scale-free network system more vulnerable. Furthermore, for fully dependent networks, heterodyne networks have the weakest ability to resist random attacks, but there is no difference between random networks and identical networks. In addition, the simulation results show that the increase of dependency will increase the vulnerability of the whole system.
【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：O157.5
本文编号：2226436
[Abstract]:With the rapid development of Internet technology and its increasingly widespread application in people's daily life, network attacks emerge in endlessly, such as bank customer information leakage, Lenovo flat price being robbed, root server, and so on. Therefore, it is very important to strengthen the network protection and build a network system with high robustness. As a powerful tool, complex networks can well describe the structural characteristics and typical dynamic behaviors of complex systems in the real world, and the attack and protection of complex networks have become a research hotspot in related fields. Under the background of complex network, this paper synthetically applies the knowledge of mathematical statistics, control, computer simulation and so on, and revolves around "how to improve the anti-attack ability of the network", "the relationship between the network anti-attack ability and the synchronization ability", and so on. The influence of clustering coefficient on the vulnerability of dependent network structure is studied. The main achievements are as follows: (1) the optimization of network anti-attack ability based on simulated annealing algorithm; based on the optimization of network anti-attack ability, a method of preserving the network structure is presented, and the optimization model is established. The optimization problem of network anti-attack ability is transformed into a single-objective combinatorial optimization problem. Then the simulated annealing algorithm is introduced to solve the single-objective combinatorial optimization problem in order to obtain the global optimal solution. Finally, the effectiveness of the optimization algorithm is verified by numerical simulation. At the same time, the variation law of important structural characteristic parameters in the process of structural optimization is also recorded and analyzed. (2) the synchronization ability of dynamic scale-free network with onion-like structure is analyzed. Firstly, the anti-attack ability of the network is optimized and observed. The optimized network topology is onion-like, that is, the nodes with large degree form the core and decrease the degree from the inner layer to the outer layer. Then, the synchronization stability of the dynamic scale-free network with onion-like structure is studied. The numerical simulation results show that the synchronization ability of the network becomes weaker with the improvement of the network anti-attack ability. At the same time, two typical structural parameters, the average path length and the degree matching coefficient, of the onion-like network are analyzed. This paper theoretically explains the reason why the ability to resist attack is improved and the synchronization ability is weakened. (3) the dependent network structure vulnerability analysis of adjustable clustering coefficient is one of the important parameters that affect the network performance. The dependent scale-free network model with adjustable clustering coefficient is proposed and a more realistic attack strategy is proposed. Considering the factors of clustering and dependency preference, the numerical simulation results show that the existence of dependent edges between the two networks weakens the anti-attack ability of the network, and the high clustering makes the dependent scale-free network system more vulnerable. Furthermore, for fully dependent networks, heterodyne networks have the weakest ability to resist random attacks, but there is no difference between random networks and identical networks. In addition, the simulation results show that the increase of dependency will increase the vulnerability of the whole system.
【学位授予单位】：天津理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：O157.5
【相似文献】
相关重要报纸文章 前1条
1 王云;如何选购UTM？[N];计算机世界;2006年
相关硕士学位论文 前2条
1 李瑞琪;复杂网络抗攻击能力优化问题的研究[D];天津理工大学;2016年
2 侯一凡;抗攻击能力评价指标体系的构建与评价方法研究[D];解放军信息工程大学;2007年
，本文编号：2226436
本文链接：https://www.wllwen.com/kejilunwen/yysx/2226436.html
