A new approach to network visualisation
The sheer size of the Internet makes it very difficult to represent graphically, even small parts of it. Yet being able to 'visualise' the net has never been more important as more and more people join the online population and attempt to navigate the unique landscape of the world wide web. It could also be exploited to help analyse network traffic data. Researchers with the Universitaet Karlsruhe in Germany took up this challenge in the framework of the COSIN project which was sponsored by the EU's Information Society Technologies Programme. The approach involved experimenting with new graph-drawing algorithms that allow the identification of specific network fingerprints, such as the hierarchical structure of computer generated and real world networks. Even though a fair amount of software was available to enable the visualisation of networks data, the state of the art at the time seemed too heuristic to be satisfactory to support network analysis. In finding a suitable tool for the 'visualisation' of very large networks (millions or even billions of nodes), the team at Universität Karlsruhe in Germany collaborated closely with other members of the COSIN research consortium. With the use of graph theory and statistical methods, they were successful in dealing with a multi-layer description of large networks that seems to reproduce most of their properties. Furthermore, once decomposed into a number of sub-systems, non-trivial and readable drawings can be established that showed all nodes and edges. In addition, existing methods of graph drawing were modified and combined in order to address the needs of specific types of networks, such as Autonomous system (AS) networks. The Universitaet Karlsruhe scientists hope to continue their work extending algorithm theory to network visualisation.
