Analysis and simulation of Voice over Internet Protocol on a campus sized computer network

This thesis proposes an analytical approach, supported with simulation results, to determine the number of Voice over IP (VoIP) sessions that can be supported by a campus-sized computer network.;Prior to the introduction of voice onto a small campus-sized network, various factors were examined. One of the most important factors affecting performance on the network is the existing background traffic on the network. Hence before deploying VoIP, the network traffic measurements should be analyzed to check for network nodes that are highly utilized.;Another important factor is the number of voice sessions that can be supported by the network. A mathematical approach is proposed to analyze this. This approach takes into account various factors in determining the number of voice sessions. One of the factors is the choice of CODEC scheme. A few of the widely used CODECs are G.711, G.729 and 0.723.1.;Other factors affecting the number of voice sessions are the end-to-end delay, packet loss, and bandwidth requirements for a single bidirectional call and finally the percentage of the utilization of the network nodes. Based on these factors the analytical approach determines the total number of voice sessions that can be supported by the network. The approach is also tested for its correctness by simulating voice on a campus-sized network using the OPNET Modeler simulation package. Three different scenarios for the same basic campus-sized network were simulated, using a different CODEC in each of them.;The simulation results validate the analytical approach in determining the number of voice sessions. It was found that the network could handle the largest number of voice sessions when the G.723.1 CODEC was used. The network supported close to the same number of voice sessions when G.711 or G.729 is used as the CODEC scheme.