基于AS间合作关系的互联网层次结构演化分析及建模

发布时间：2018-10-29 21:30

【摘要】：随着互联网经济的快速发展,互联网已然成为带动国民经济发展的新引擎,如何更快、更好的为更多用户提供互联网信息服务成为互联网基础服务提供商面临的关键问题,因此对于研究互联网服务提供商在商业利益和需求的驱动下如何建立网络之间的关系对于理解互联网拓扑结构、性能、演化和动态性具有重要的意义。首先,分析了互联网拓扑中AS间的合作关系。本文提出了基于节点介数中心性指标的AS间关系推断算法,该算法基于派系网络、节点介数和AS路径三元组等方面推断出AS间的P2C和P2P关系,可避免CAIDA算法对于P端传输度小于C端传输度推断失误的情况,且其计算过程优于CAIDA算法。经过真实的AS关系数据验证,基于节点介数中心性指标的AS间关系推断算法的准确率：P2C为99.5%,P2P为99.1%,同时由于本算法和CAIDA算法均采用了派系网络和AS路径三元组,所以相比GAO、XIA、UCLA等算法在总体的准确率要高。然后,研究了AS级互联网拓扑结构的层次性。首先,从AS级网络的整体演化趋势上出发,观察网络节点和边的数量演化情况,并根据不同区域AS注册数量研究AS级互联网宏观拓扑结构各地区的增长趋势,发现北美和欧洲地区的网络发展主导了当前互联网的增长。其次,从网络结构角度出发研究了网络结构熵、网络深度、网络核和网络集聚性等刻画网络层次性的特征量,分析了AS级互联网拓扑结构的层次性演化,结果表明AS级互联网拓扑结构在向扁平化层次结构方向发展。再次,构造以派系网络节点为开始,由P2C关系为连接的自上而下的锥形层次结构。最后,从派系网络、节点层次分布、P2C/P2P层次连接偏好、节点度连接偏好等方面分析了锥形层次结构,由四个方面的演化结果得到如下结论：互联网存在稳定的派系网络；节点数量的层次分布服从正偏态分布,且其层次越低,节点最大度值越小；P2C偏向于两个锥形层次结构间的高层连接,P2P偏向锥形层次结构内部的跨层连接；P2C/P2P连接的节点度分布服从幂律分布。最后,提出层次连接偏好建模方法。基于互联网拓扑结构的节点和边动态增长、层次结构、层次和节点度偏好连接以及P2C/P2P商业关系等特性,提出了基于层次性偏好连接的互联网建模方法HPA。经过大量网络仿真实验和结果分析,发现HPA网络模型很好的体现了互联网的动态性、层次性、偏好性和商业关系。
[Abstract]:With the rapid development of the Internet economy, the Internet has become a new engine to drive the development of the national economy. How to provide Internet information services for more users quickly and better has become a key problem faced by Internet basic service providers. Therefore, it is of great significance to study the relationship between Internet service providers driven by business interests and needs for understanding the topology, performance, evolution and dynamics of the Internet. Firstly, the cooperative relationship between AS in Internet topology is analyzed. In this paper, we propose an algorithm for inferring the relationship between AS based on the index of node centricity. The algorithm inferred the P2C and P2P relationships between AS based on factional network, node meshes and AS path triples. The CAIDA algorithm can avoid the error of P terminal transmission degree less than C terminal transmission degree inference, and its calculation process is better than CAIDA algorithm. Based on the real AS relation data, the accuracy of the AS interrelation inference algorithm based on the central index of node number is as follows: P2C is 99.5 and P2P is 99.1. At the same time, the algorithm and the CAIDA algorithm both use the factional network and AS path triples, so the overall accuracy of the algorithm is higher than that of the GAO,XIA,UCLA algorithm. Then, the hierarchy of AS level Internet topology is studied. First of all, based on the overall evolution trend of AS level network, the evolution of network nodes and edges is observed, and the growth trend of each region of AS level Internet macro topology structure is studied according to the number of AS registrations in different regions. Found that the development of the network in North America and Europe led to the current growth of the Internet. Secondly, from the point of view of network structure, the characteristics of network entropy, network depth, network core and network agglomeration are studied, and the hierarchical evolution of AS network topology is analyzed. The results show that the AS level Internet topology is developing towards flat hierarchy. Thirdly, we construct a top-down conical hierarchy, which starts with factional network nodes and is connected by P2C relation. Finally, the conical hierarchical structure is analyzed from factional network, node hierarchical distribution, P2C/P2P hierarchical connection preference, node degree connectivity preference and so on. The following conclusions are obtained from the evolution results of four aspects: there is a stable factional network on the Internet; The hierarchical distribution of the number of nodes follows the positive skew distribution, and the lower the level is, the smaller the maximum degree of the node is; P2C is inclined to the high-level connection between the two conical hierarchies, and P2P to the cross-layer connection within the conical hierarchy. The node degree distribution of P2C/P2P connections follows the power law distribution. Finally, a hierarchical connection preference modeling method is proposed. Based on the characteristics of node and edge dynamic growth, hierarchical structure, hierarchical and node degree preference connection and P2C/P2P business relationship, a hierarchical preference connection based on hierarchical preference connection (HPA.) is proposed. Through a large number of network simulation experiments and results analysis, it is found that the HPA network model well reflects the dynamic, hierarchical, preference and commercial relationship of the Internet.
【学位授予单位】：东北大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02

【相似文献】