A Methodology for Magnitude-based Inter-AS Distance Estimation

The relationships among autonomous systems (ASes) can be categorized into three types: 1) transit, 2) peering and 3) sibling. Since customer ASes purchase the Internet access from provider ASes over transit links and any two ASes can exchange their traffic without any payment over peering links, transit links cost more than peering links from the economical point of view. We propose an inter-AS distance estimation method with eigenvalue analysis based on a traffic-flow/volume assumption so that transit links are more distant than peering links. We evaluate this method with two topology datasets: 1) the topology generated by a topology generator Inet-3.0 and the inferred Internet topology from the real Internet routing table. We show that the proposed method can estimate the inter-AS distance on the basis of types of AS relationships (i.e., transit is more distant and peering is closer).