Research On Dynamic Layout And Visual Analysis Of Network Communication Data

Network communication data not only reflects the relationship between network users,but also has the characteristics of dynamic behavior and high dimension.At the same time,it contains the user's behavior patterns and usage patterns in the process of surfing the Internet.In comparison to the traditional data analysis methods,visualization and visual analysis technique is intuitive and interactive,allowing users to intuitively perceive the network structure and network node relationship combining explore the data interactively,getting immediate feedback and analysis without any prior knowledge.In addition,it is a beneficial supplement to the traditional data analysis method.In this thesis,we mainly do research about the traffic attribute,time-varying characteristics and interaction design of network communication data.Our work deepens user's understanding of network communication data and assists the user to make effective analysis.The main points of our research work are as following:1.A force-directed layout algorithm based on traffic flow.In the communication network,the network communication implies a lot of information.The flow of a single node reflects its important degree and the active degree in the network.The traffic flow between multi-nodes reflects the inter node communication relationship and closeness degree.The traditional network layout algorithm can only reflect the basic topology of the network,but cannot reflect the communication between nodes and attributes.In this thesis,we proposed a force-directed layout algorithm based on traffic flow,which not only establish importance index according to the traffic flow,but also improved the traditional FR algorithm.Our algorithm reflects the traffic information and connection relationship between nodes in the layout.2.A triangular mesh-based algorithm of dynamic layout.Communication network has time-varying characteristics,change of traffic information in the network makes the connection relationship between nodes change with time,which leads to the dynamic change of network structure.In order to ensure the relative stability of the layout,the layout changes gradually rather than sharply,we proposed a triangular mesh-based algorithm of dynamic layout into which nodes and connection relationship in the network are mapped.In order to maintain the stability of the metal map,we control the node's position change according the importance degree of the triangle mesh.Experiments show that this method can effectively control the movement of important nodes,and thus keep the intention graph of the network.3.An interaction strategy based on enhanced brush operation.Except for the topology information,network communication data contains a variety attributes,such as traffic flow,port,IP address,etc.Network communication data has high dimension features.Parallel coordinates can effectively display high dimensional data.The brush operation can be effective for data filtering.However,when it comes to big data,users often have difficulty in selecting operation.In this thesis,and enhanced brush operation strategy is proposed to increase the distribution of data in the process of interaction,improve the interactive efficiency and is more effective to assist users to explore the data.