Inferring Deployment Strategies of Internet Service Providers Based on Geographical Coverage-Area

Internet serves as the underlying basis for knowledge grid. The growth of AS-level Internet topology results from not only customer choices but also provider strategies. This paper focuses on inferring the un-highlighted AS provider deployment strategies based on the mining of AS geographical coverage properties. Taking an AS to be an area covered by the cities its routers reside in and the registered cities of the AS and its neighbors to be its approximate router locations, the data-mining is feasible with publicized topology measurement and administration data. The pan-coverage-area (PCA) and the coverage-area (CA) of an AS are defined to characterize the AS connectivity and service coverage respectively. Via association analysis, AS coverage and service deployment strategies are inferred and classified. The CA sizes are found more relevant to business strategies than topological metrics. This work is helpful for understanding and modeling the practical origins of the Internet evolution.