Evaluation of Optimal Routing Policies in Communication Nets by Simulation

A Monte Carlo simulation program that enables the evaluation of different routing policies in a communication network is described. This simulation is intended to deal with nonoriented (telephone) networks and the grade of service (G/S) is taken as the performance measure. However, it can easily be adapted to deal with oriented or mixed networks. The program assigns to each branch of the analyzed network a price function. For each call, the route of minimal cost, among all available under the present load conditions in the network, is then chosen. The program was used for comparing routing policies, as initiated by different price functions. The comparisons were made on several test networks and for different traffic loads. The results obtained showed a significant improvement in the G/S for the policy that aimed at distributing the flow through the network in the most homogenous way possible. This homogenous flow distribution was obtained by using the above minimal-cost policy and a price function, assigned to each channel, that monotonically increases with the traffic load of the branch to which the particular channel belongs. The influence of the topological configuration of the network on the G/S, under equal external demands, was also investigated and the results are given. The way in which the results are printed out enables the simulation program to serve also as a practical and useful tool in the design (or the redesign) of a communication net.