Research On Visualization Method Of Space-based Networks Based On Time-varying Characteristic

In the future information war, relying on the powerful spatial coverage of space-based network(SBN) to realize information acquisition, transmission, processing and distribution, is the key aspect to achieve the information superiority. The topology of SBN has apparent time-varying characteristic, in the early stages of the network construction, there are a few nodes and links, hence, it don’t affect the commanders to understand them. However, along with the network scale is more and more big, and the structure of the network is more and more complex, leading to commander is hard to get visually and intuitively understanding of time-varying regularity of SBN and make accurate judgment and scientific decisions. While the visualization technique can show the invisible data as computer graphics and assist commander to intuitively perceive SBN, it is also an open problem for time-varying network visualization. Moreover, the existing visualization methods of SBN focus on simulating the physical layer and link layer, but lacking the visualization of network layer and application layer. Therefore, it is quite a challenging work for researching the visualization methods of SBN based on time-varying characteristic.Around the time-varying characteristic of SBN, this thesis researches the visualization framework, visualization layout and visual analytics, and makes some innovative research results. The main jobs and innovations of the dissertation are as follows:Firstly, to design a visualization framework based on time-varying characteristics of SBN. Design visualization method needs systematic thinking, however, the present research is still in its infancy and lacks a mature and systematic framework. This dissertation deeply analyzes the time-varying characteristics of SBN topology, and elaborates the SBN visualization tasks based on multilevel nest model; on this basis, this dissertation presents a SBN visualization framework based on time-varying characteristics called SBNVis and systematically designs the process of SBN visualization. Under the guidance of SBNVis, we can directly and systematically achieve the visualization of SBN.Secondly, to propose a trellis layout approach combining temporal and spatial context. The time-varying characteristics of SBN hide in the time series data, through comparing the layout of neighboring timeslices, we can find the similarity and dissimilarity, and understand hidden rules of dynamic change. However, the present layout approaches have two drawbacks, that is, low-quality layout of the single timeslice, and unstable layout of timeslice sequence. And for this, this dissertation proposes a hybrid force directed layout(H-FDL) algorithm to optimize the layout of single timeslice, and a combining temporal and spatial context layout approach to improve the stability of layout of timeslice sequence. Experiments have shown that, compared with traditional methods, some indicators are optimized, such as the number of edge cross, the average angle of crossed edges, the average length of edges, angular resolution, and so on. The statistical indicator of movements’ distance of nodes in neighboring timeslices is improved.Thirdly, to propose an animation layout approach based on difference map event driven. Existing animation layout approaches regard timeslices sequence as keyframes sequence of animation, and utilize the interpolation frames to complete the transition; substantially they create local interval animation, rather than global continuous animation, hence, the animation is inflexible and unsmooth, so it is hard to show the time-varying characteristic of SBN. For this, we propose an animation layout approach with difference map event-driven. Firstly, we divide the whole animation process into two states, i.e., steady state and fluctuation state. Then, for the latter, we propose a staged animation transition strategy, to reduce the out-of-order movements of nodes, thereby, to resolve the visual clutter problem caused by drastic fluctuation. Thirdly, for steady state, we show the smooth layout of network under force-direct algorithm directly. Experiments have shown that, compared with traditional animation layout methods, we gain better effects with smooth layout, layout quality, being distinguishable, controllability and visual independence.Fourthly, to propose interactive analysis approaches based on visual interface for representative visual analytics requirements. Firstly, for the requirement of exploring hidden pattern, we propose an interactive layout method based on implicit links, which enables user to explore semantic cluster based on force directed layout and improves the interaction. Then, for the requirement of detailed information query, we propose a visual query method based on magnifier metaphor and build query views. It gain better effects on consistency, decomposable visual elements and magnify local view.Lastly, to design and implement a prototype system of SBN called SBNVis Sys, which verifies the visualization approaches and key technologies, and is applied to related projects.