Inferring Internet Topology at Point of Presence Level

Analyzing point of presence (PoP) level Internet topology offers several advantages for researchers, network operators, and those involved in optimizing and maintaining the Internet infrastructure. It provides insights into how the Internet backbone is organized and how various networks are connected at a fundamental level. Specifically, understanding PoP level Internet topology is essential for optimizing network performance, ensuring reliability, and developing protocols that facilitate seamless communication across the interconnected web of networks on a global scale. PoP maps allow us to go beyond the simple autonomous system (AS) level abstraction by supporting multiple connections among the ASes. In this study, we propose a set of techniques to infer the PoP level topology map of the Internet by utilizing several real-life datasets, including traceroute, DNS, and commercial geolocation databases. Initially, we map the IP addresses collected from the traceroute to their corresponding ASes. Then, we use six geolocation techniques to find the location of IP addresses. Subsequently, we group the IP addresses into PoP nodes based on their geographical locations within ASes and identify the connection between PoP nodes. We validate this approach using various research and commercial networks worldwide. We investigate various features of the map to provide more insight regarding the backbone structure of the Internet.