GENESIS: An agent-based model of interdomain network formation, traffic flow and economics Aemen Lodhi Amogh Dhamdhere Constantine Dovrolis School of Computer Science CAIDA School of Computer Science Georgia Institute of Technology University of California San Diego Georgia Institute of Technology
[email protected] [email protected] [email protected] Abstract—We propose an agent-based network formation however, can have a global impact on the economic viability model for the Internet at the Autonomous System (AS) level. of all ASes and the structure of the Internet. The proposed model, called GENESIS, is based on realistic The Internet remains in a persistent state of flux subject to provider and peering strategies, with ASes acting in a myopic and decentralized manner to optimize a cost-related fitness function. changes in various exogenous factors. How will the Internet GENESIS captures key factors that affect the network formation change due to consolidation of content [1], large penetra- dynamics: highly skewed traffic matrix, policy-based routing, ge- tion of video streaming, falling transit prices [2], expanding ographic co-location constraints, and the costs of transit/peering geographic footprint of content providers [3], cheap local agreements. As opposed to analytical game-theoretic models, availability of peering infrastructure at IXPs [4]? We propose which focus on proving the existence of equilibria, GENESIS is a computational model that simulates the network formation a computational agent-based network formation model, called process and allows us to actually compute distinct equilibria (i.e., “GENESIS”, as a tool to study such questions. GENESIS is networks) and to also examine the behavior of sample paths modular and easily extensible, allowing researchers to exper- that do not converge.