Poster Detail View larger image. (~115K)

The dataset contains the packet counts, by two-hour period, between each pair of countries. Each country is represented by a box- shaped glyph (graphical object) that is scaled and colored to encode the total packet count for all links emanating from the country. The glyphs are positioned at the locations of the countries' capitals and extend perpendicular to the surface of the globe. The color-coded arcs between the countries show the inter- country traffic, with the higher and redder arcs indicating the larger traffic flows. The globe is illuminated by a light which is positioned to indicate, via the angle of the sun, the time for the frame of the time-series data that is displayed.

The image was produced at Bell Laboratories, whose interest and motivation for visualizing graphs comes from analyzing many of their associated networks. In this usage, a network is a graph where the associated statistics represent traffic. The traffic may represent a time-varying statistic, as with the call flow in a telecommunication system throughout the day; a constant, as with network capacity; or a stochastic statistic, as with the number of IP packets sent between routers on a backbone data network.

The image was generated with "SeeNet3D," a Bell Labs network visualization environment for the exploration of large, time- varying communication networks. The research focus is on how to overcome some fundamental problems in understanding large and complex networks using 3D graphics technologies, while simultaneously maintaining the benefits of the useful and well- established 2D node and link maps.

The SeeNet3D system is currently a 5,000 line C++ program built on top of the Vz framework. Vz is a visualization platform embodied in an object-oriented, cross-platform (MS Windows, OpenGL, and X11) C++ library.