The Design And Implementation Of Network Flow Data Visual Analysis System

Network flow data is a large number of network log data collected by monitoring equipment.As an important carrier of network events,network flow data is important basis to judge the network running status for analysts.Because network flow data has many dimensions and analysis levels,and the data is huge,so using the traditional analysis methods is difficult to carry on the comprehensive and detailed exploration,at the same time bring heavy cognitive load to analysts.Visual analysis deals with the data by means of visualization,in combination with user interaction that analysts can quickly mining valuable information from huge amounts of data,so analysts can make effective decisions.Currently,the visual analysis of network flow data need to solve the following problems: on the one hand,aiming at network flow data's characteristics that multi-level and the correlation of hosts,the thesis want to put forward effective analysis method.Existing methods focus on a single level,difficult to achieve multi-level analysis.On the other hand,according to the characteristics of the analysis object and multivariate time-varying of network flow data,the system design reasonable views and layout methods,so as to realize the high efficiency of visual analysis.This thesis study the visual analysis methods of network flow data in view of the above two problems,and the research results are as following.1.The multi-level correlation visual analysis model is designed for the characteristics that multilayered and correlation of hosts,and it realizes the visual analysis from the overall to local,and from local to the individual.The model including the function that data conversion,visual analysis,and views generation.Visual analysis is the core of this model.Visual analysis is divided into multi-level analysis module and correlation analysis module.Set in the characteristic of multilayered,Multi-level analysis designs the macro level that analyzes whole time series,the medium level analyzes local time series,and the micro level analyzes particular subject.Multi-level analysis realizes the analysis process that from global to local,and from local to individual.Set in the characteristic of the correlation of hosts,correlation analysis builds the relationship network between the hosts,analyzes the correlation between the nodes of hosts,gets germane nodes,and realizes the process from node to plane.2.Visual encoding,views and layouts are designed and realized according to the above model and the characteristics of multivariate time-varying of the network flow.Multi-level correlation visual analysis model provides analysis object,analysis process and target output that a visual design needs.At the same time,the network flow data of multivariate time-varying characteristics should be combined with the classical multivariate time-varying data visualization methods.Based on the above two points,this thesis improves and designs visual encoding,views and layouts of the network flow data visualization,completes the task analysis effectively,and realizes visualization expression of multi-level correlation visual analysis model and multivariate time-varying data.3.Based on the above method,this thesis realizes the network flow data visual analysis system.This system implements visual analysis of network flow data by combining rich interactive functions.This system uses the authoritative data set of the VAST challenge for case studying,through multi-level analysis and correlation analysis realizing the experimental analysis of the related cases,and proving the validity and practicability of this system.