Scalability Analysis of an Emulated Platform for Heterogeneous Networks

Abstract

In the world of everyday changing wireless communication systems, it is difficult to extend hardware-based test-beds which can reflect constant changes in the technology. Emulation, which is an alternative method to evaluate the services in realtime, lacks in accuracy as the scale of the network increases. In this paper, we present the scalability analysis of the open-source discrete event emulator NS-3 and demonstrate its limitations under different load conditions. We, further, model the underlying radio network as an open GI/G/1 queuing network with nonexponential service time and inter-arrival time distributions to present theoretical analysis of the system. Theoretical results give a trustworthiness indicator and provide a measure of divergence between emulated and analytical results. We use Percentage of Accuracy (PoA) as the measuring parameter of accuracy of emulation results with respect to theoretical results. We also conduct a small test-bed based experiment and compare the results of the same with that of the emulator. Test-bed results help us to tune the parameters of the emulator, so that better emulation accuracy can be obtained.