Power laws is a relatively new paradigm used to model networks. Instead of the traditional concept of a graph and the degree of a node (number of nodes, number of edges, and diameter), power laws defines the characteristics of the network components as an expression of the form “y is proportional to xª,” with ª being a constant, and x and y measures of interest. For example, “The degree of a node v is proportional to its rank to the power of a constant R.”
The authors use the power laws modeling concept to define and predict Internet growth. They use actual data, collected by two different sources, to demonstrate their model’s correctness at the autonomous system level (that is, Internet subnets that are under separate administrative authorities).
Power laws modeling seems robust enough to deal with otherwise highly skewed data, and this makes it a reasonable tool to be used in Internet simulations.