A method of tree network topology inference based on hierarchical host table

Abstract

Network topology inference uses network measurement methods to find out the elements of internet in the target field and their connection relationship. In traditional network topology inference, traceroute-like methods which bases on the feedback information of routers are used, but anonymous routers seriously affect the performance. So tomography relying on end-to-end measurements has become a hot topic. This paper provides a method of topology inference based on hierarchical host table. On one hand, this method doesn't require any cooperation from the internal routers, which can avoid the problem of anonymous routers; on the other hand, unlike traditional tomography techniques, we limit the clustering problem on the same router level, which not only improves the topology inference accuracy, but also reduces the measurement cost. The simulation on NS-2 shows this method can get the logical network topology effectively.