On inferring regional AS topologies

Analysis of AS-level topologies is an effective way to understand the Internet structure and its development. An AS topology represents logical connection relationships of networks, and thus, is independent of geographic locations of connections. However, the actual Internet topologies are constructed with geographical and physical limitations. In the meantime the global Internet topology is shifting from a U.S. centered star topology to a more distributed topology interconnecting regional hub ASes. Therefore, regional topology views would provide new insights to understanding changes in the Internet structures. Our goal is to understand differences and similarities a-mongst Internet structures in regional views such as from Europe and from Asia. We examine inference method for regional AS topologies, and compare the resulting regional AS topologies. Our inference technique extracts AS boundaries in traceroute data collected from multiple vantage points. Then, we illustrate the resulting regional AS topologies using AS Core Maps in order to identify the degree structure and AS locations in each regional view. We show that the distributions of AS out-degrees similarly follow power-law but Tier-1 ASes play different roles in regions. We also identify limitations such as the lack of vantage points in Africa and Oceania in the current data sets.