Computing Observed Autonomous System Relationships in the Internet

Autonomous Systems (ASes) in the Internet use BGP to perform interdomain routing. BGP routing policies are mainly determined by the business relationships between neighboring ASes, which can be classified into three types: provider-to-customer, peer-to-peer, and sibling-to-sibling. ASes usually do not export provider routes and peer routes to providers or peers. It has been proved that if all ASes conform to this common export policy then all AS paths are valley-free. Since AS relationships are not publicly available, several studies have proposed heuristic algorithms for inferring AS relationships using publicly available BGP data. Most of these algorithms rely on the valley-free property of AS paths. However, not all AS paths are valley-free because some ASes do not conform to the common export policy. As a result, inferred AS relationship are inaccurate. Instead of inferring AS relationships, we propose an algorithm for computing observed AS relationships based on transit relationships between ASes that are revealed by BGP data. We analyze the types of mismatches between observed AS relationships and actual AS relationships and show that the mismatches can be used to identify ASes that violate the common export policy.