OSPF protocol and statistical tools for network simulations in NS-2

Abstract

Network simulations have become a popular method to study network behavior. However, relevant and consistent results can be obtained only if the model has sufficient details and the simulation and analysis procedures meet high enough quality criteria. The solution space of simulations may have several clearly separate, but equally probable optimal minima. This together with the stochastic nature of simulations typically necessitates several long simulations. Here we describe implementations of the OSPF protocol and a novel random generator in NS-2 simulator. Additionally, we describe a tool that enables efficient statistical analysis of the simulation results. The power of these tools is demonstrated with a simple scenario.