This article provides a comparison among different methods for estimating the aggregation of Internet traffic resulting from different users, network-access types and corresponding services. Some approximate models usually used as isolated methods are combined with a temporally scaled ON-OFF model with binomial approximations. The aggregation problem is solved using a new form of parameterization based on the composition of the source traffic accordingly to the concrete characteristics of the users, the accesses and the services. This is a new concept, called CASUAL, included within an overall network planning methodology for the design and dimensioning of Next Generation Internet.

network planning, traffic modeling, network scenario