Some results on estimation and modeling of switch transit traffic in a backbone network

Estimating the transit traffic via a node in a communications backbone network such as ATM is an important aspect of the capacity planning in a network. In this paper, we propose a graph theoretic approach to estimate the transit traffic and present results of simulation of switch traffic in a network. Modeling is done using graph theoretic methods and traffic on a trunk is estimated using an updated version of the Bellman-Floyd algorithm incorporating the optimal policy matrix derivation. The simulations are run for various random distances, random connectivity between nodes and random traffic patterns. We show that the average ratio of the transit to the originating traffic at any node is directly proportional to the average number of hops in the routes, which in turn is directly proportional to the average degree (average number of trunks at any node) of a node in a network. The simulation results indicate that the transit traffic may be well above one hundred percent of the originating traffic, and this increases with the average ratio of the nodes to the degree of the node