An incremental super-linear preferential internet topology model extended abstract

By now it is well known that the distribution of node degrees in the graph induced by the peering arrangements between Autonomous Systems (ASs) exhibits power laws. The most appealing mathematical model that attempts to explain the power-law degree distribution was suggested by Barabási and Albert (the BA model). We introduce two new models that are extensions to the BA model: the "Incremental Edge Addition" (InEd) model, and the "Super-Linear Preferential Attachment" (SLiP) model. We prove that both our models are more successful in matching the power-law exponent, in producing leaves, and in producing a large dense core. Beyond mathematical analysis, we have also implemented our models as a synthetic network generator we call TANG (Tel Aviv Network Generator). Experimentation with TANG shows that the networks it produces are more realistic than those generated by other network generators.