网络通信数据可视化分析方法研究

发布时间：2018-04-26 16:51

本文选题：网络通信数据 + 信息可视化　；参考：《国防科学技术大学》2014年硕士论文

【摘要】：伴随着社会和科学的快速发展,计算机网络已经成为成为了几乎覆盖世界所有角落的庞大系统。伴随这计算机互联网规模的日益庞大,其每时每刻所产生的网络通信数据也成几何爆炸式的增长,直接推动社会进入大数据时代。而计算机网络通信数据在许多领域都有着不可替代的作用。然而,在现阶段对网络通信数据的利用并不是很理想,还存在着实时性差、误解率高的问题。而可视化分析技术的日益成熟,其强大的辅助分析能力体现的越来越充分。所以,对网络通信数据进行可视化分析是一个非常有意义的研究工作。针对网络通信数据的特点,以及在不同应用背景下用户对网络通信数据分析的不同需求,本文研究设计了多种可视化方案和相应的交互策略,在提高用户对繁杂网络通信数据的认知能力和分析能力。本文的研究工作主要有以下几点:(1)针对用户对网络节点之间的通信可达性关系分析的需求,提出了一种基于通信强度集束的平行坐标系可视化方法,通过定义可达性矩阵、通信强度矩阵,在统一视图下直观表现感兴趣节点、源节点、目标节点集合及其分布,以及它们之间的可达关系,以适合用户理解的强、中、弱三级概念对可达关系曲线进行模糊集束,并辅以适合的人机交互手段,达到局部放大、快速筛选的目的。(2)针对用户对某些特殊的网络通信行为分析的需求,总结出这些通信行为固有的通信模式及其特点,设计了基于平行坐标系的可视化方案,利用不同类型通信行为表现出的视觉特征,以及人眼的视觉发现能力,辅以人机交互手段,使用户可以对特殊区域进行筛选,提高其从繁杂的网络通信流中发现某种特殊的网络通信行为的能力。(3)常规的网络拓扑图仅能表现节点及节点间的关系,难以支持后续的网络业务结构分析等需求,本文提出了一种流量力引导布局算法,根据流量信息形成带有节点聚类的网络节点图,不仅能够展现节点间的连接关系,而且能够体现这种关系的流量属性以及节点的类属特征,通过交互手段,还可以辅助实现多粒度分析、活跃节点发现等功能。(4)针对用户在多层网络结构中对网络通信流承载关系分析的需求,提出基于层次布局的网络通信流可视化方案。针对计算机网络不同布局方式之间所具有的对应关系和差异性,通过对已有布局的局部调整,在统一的视图下清晰展现多层结构及其映射关系,辅助用户对网络通信流承载关系的可视化分析。
[Abstract]:With the rapid development of society and science, computer network has become a huge system covering almost every corner of the world. With the increasing scale of the computer Internet, the network communication data generated by the computer at every moment has also become a geometric explosive growth, directly promoting the society into the big data era. The computer network communication data plays an irreplaceable role in many fields. However, at present, the utilization of network communication data is not ideal, and there are still some problems such as poor real time and high misunderstanding rate. With the development of visual analysis technology, its powerful ability of auxiliary analysis is becoming more and more perfect. Therefore, the visual analysis of network communication data is a very meaningful research work. In view of the characteristics of network communication data and the different requirements of users for network communication data analysis under different application backgrounds, this paper studies and designs various visualization schemes and corresponding interactive strategies. In improving the users of complex network communication data cognitive and analytical ability. The main research work of this paper is as follows: 1) aiming at the demand of users to analyze the communication reachability relationship between network nodes, a parallel coordinate system visualization method based on communication intensity cluster is proposed, and the reachability matrix is defined. The communication intensity matrix, which directly represents the set of nodes of interest, source nodes, target nodes and their distribution in a unified view, and their reachability, in order to suit the user's understanding of the strong, medium, The concept of weak three level carries on the fuzzy cluster to the reachable relation curve, and adjusts with the suitable man-machine interactive means, achieves the local magnification, the fast sieving goal. (2) according to the user's demand for some special network communication behavior analysis, This paper summarizes the inherent communication modes and their characteristics of these communication behaviors, designs a visualization scheme based on parallel coordinate system, utilizes the visual characteristics of different types of communication behaviors, and the visual discovery ability of human eyes. By means of man-machine interaction, users can screen special areas and improve their ability to find some special network communication behavior from the complicated network traffic.) the conventional network topology can only show the relationship between nodes and nodes. It is difficult to support the requirements of subsequent network service structure analysis. In this paper, a traffic force guided layout algorithm is proposed. According to the traffic information, a network node graph with node clustering is formed, which can not only show the connection relationship between nodes. Moreover, it can reflect the flow attribute of this relationship and the generic characteristics of nodes. By means of interaction, it can also help to realize multi-granularity analysis. Active Node Discovery and other functions. (4) aiming at the demand of users to analyze the bearing relationship of network traffic in multi-layer network structure, a network traffic visualization scheme based on hierarchical layout is proposed. In view of the correspondence and difference between different layout modes of computer network, the multi-layer structure and its mapping relationship are clearly displayed under the unified view by local adjustment to the existing layout. Visual analysis of network traffic load-bearing relationship by auxiliary users.
【学位授予单位】：国防科学技术大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.0

【参考文献】