复杂网络多层次差异性建模及其应用研究

发布时间：2018-08-03 11:49

【摘要】：近几年互联网、物联网、大数据的快速发展,使得人与人之间、人与物之间、物与物之间的距离越来越近。网络科学又重新进入专家学者的研究范围内。然而因为现实生活的复杂性,来源于具体事物的抽象网络也具有各自的特点,这具体表现在节点异构、链路异构、网络拓扑异构等方面。目前复杂网络的经典模型(ER随机图、小世界模型、WS无标度模型),虽然能模拟现实生活中的部分网络,但是由于现实生活的复杂化越来越大,这些经典模型的局限性也越来越明显。所以如何构造一个能适用于现实生活中绝大多数网络的网络模型成为网络科学领域研究的热点。针对以上问题,本文从网络拓扑结构出发,忽略网络中节点或链路在现实生活中的异构性,提出了一种构造多层次差异性网络的建模方法,并将这种建模方法应用于混合拓扑结构网络和混合模型网络。本文的创新点和主要的内容如下:1、提出一种新的建模方法,即多层次差异性建模。其中多层次主要体现在利用k-壳分解方法挖掘网络的多层次结构并将其构造成为一个主从多机通信系统,差异性主要体现在根据网络中节点和链路的自身属性和整个网络的拓扑结构分别计算其重要度。2、针对混合拓扑结构网络进行建模,以航天信息系统为例。首先构造航天信息系统的主从多机通信机制;然后根据网络的拓扑结构计算了节点重要度和链路重要度;接下来还进一步根据网络中具体的消息传递情况计算了节点重要度和链路重要度。3、针对混合模型网络进行建模,以美国电力网络为例。首先基于k-壳分解方法将整个网络构造成为一个主从多机通信系统,形成多层次网络;然后根据网络的拓扑结构计算了节点重要度和链路重要度,并且与各种基本指标进行了对比;最后分析了本文所建立的电力网络模型的脆弱性与鲁棒性。通过以上研究本文实现了复杂网络多层次差异性建模,这种方法可忽略实际网络中节点或链路的异构性,可对拓扑结构复杂的网络进行分析,且该方法还可根据网络的混合模型构造多层次网络。由对混合拓扑结构网络和混合模型网络的仿真结果可知,这种建模方法具有普适性,并且能识别出重要节点或链路,通过保护这些节点或链路可降低网络的脆弱性。
[Abstract]:In recent years, with the rapid development of Internet of things and big data, the distance between people, things and things is getting closer and closer. The network science again enters the expert scholar's research scope. However, because of the complexity of real life, abstract networks from specific things also have their own characteristics, which are embodied in the heterogeneous nodes, heterogeneous links, heterogeneous network topology, and so on. At present, the classical model of complex networks (ER random graph, small world model / WS scale free model) can simulate some networks in real life, but because of the increasing complexity of real life, The limitations of these classical models are becoming more and more obvious. Therefore, how to construct a network model suitable for most networks in real life has become a hot topic in the field of network science. In view of the above problems, this paper presents a modeling method for constructing multi-level and differential networks by considering the network topology structure and ignoring the heterogeneity of nodes or links in the network in real life. The modeling method is applied to the hybrid topology network and the hybrid model network. The innovations and main contents of this paper are as follows: 1. A new modeling method called multilevel difference modeling is proposed. The multi-level structure is mainly reflected in mining the multi-level structure of the network by using the k-shell decomposition method and constructing it into a master-slave multi-computer communication system. The difference is mainly reflected in the calculation of the importance of nodes and links and the topology structure of the whole network, and the modeling of the hybrid topology network, taking the space information system as an example. Firstly, the master-slave communication mechanism of aerospace information system is constructed, and the node importance and link importance are calculated according to the topology of the network. Then, the node importance and link importance. 3 are calculated according to the specific message passing in the network. The hybrid model network is modeled, and the US power network is taken as an example. Firstly, the whole network is constructed into a master-slave multi-computer communication system based on the k-shell decomposition method, and then the node importance and link importance are calculated according to the topology of the network. Finally, the vulnerability and robustness of the power network model established in this paper are analyzed. Through the above research, the multi-level difference modeling of complex network is realized. This method can ignore the heterogeneity of nodes or links in the actual network, and can analyze the network with complex topology. The method can also be used to construct multi-level networks according to the hybrid model of the networks. The simulation results of the hybrid topology network and the hybrid model network show that this modeling method is universal and can identify important nodes or links. The vulnerability of the network can be reduced by protecting these nodes or links.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：O157.5

【相似文献】