Abstract:
Due to the proliferation of CDNs with edge networks, the known hierarchical model of the Internet is gradually changing. Very recently, it has been demonstrated that CDNs have become part of the Internet core [5]. There are various attempts by researchers to map the ecosystem of big giants like Google, Akamai, and Net
ix. However, very little attention has been given to how a regular Internet client perceives from the CDN hosted websites. This had a direct impact on anti-censorship approaches which rely on web content for their success full operation. In this research project, we rst quantify that for a regular Internet user what fraction of Alexa top 1000 websites are located inside/outside its own country. We observe that sign cant fraction of websites are hosted within the nation. For example, > 80% Alexa top 1000 websites
are located inside Saudi Arabia. We used a popular active geolocation technique, multilateration, for locating the websites. To do so, we implemented multilateralism with some Modi cations suited for our research goals. The general principle behind multilateration is to first nd delay between a target node (which needs to geolocate) and a set of reference nodes (whose locations are known apriori). Later, these delay measurements are converted to approximate distances between the target and the
reference nodes. Keeping refernce nodes as centers and distances as radii, disks are drawn on the earth's map, possibly yielding an intersection area, having the target inside it.
