Data-driven large scale network-layer Internet simulation

Abstract

Internet is a spontaneously growing complex system whose large scale structure is affected by many interacting units aimed at optimizing local communication efficiency without a central authority. Very large number of nodes; wide-spread geographical distribution; predominant role of wireless access; mobility; strong presence of internet enabled smart devices, and heterogeneity increases the complexity of the internet tremendously. This immense global entity has not been precisely characterized, even though the building blocks of the Internet as well as the protocols and individual components have been subject to intensive studies for more than two decades. In the absence of accurate maps, researchers rely on a general strategy that consists of acquiring local views of the network from several vantage points and merging these views. Such local views are obtained by measuring a certain number of paths to different destinations, through the use of probes or the analysis of routing tables. Size of the output data after running the simulations on these several vantage points can reach up to terabytes in each simulation. The merging of several of these views provides a sample of a map of the internet [1].