Predictive performance modeling and simulation

Abstract

The purpose of this thesis was to create and simulate a model of an existing Campus network with a view to predict future performance. This thesis also suggests ways in which such a network can be optimized. Such simulation and modeling can be referred to as "Predictive Performance Modeling'. In this research a model of Florida International University (University Park campus) High Speed Network was created. Simulation of the model was carried out and an ATM Backbone Analysis was done. The results obtained were compared with corresponding results obtained by network performance monitoring and measurement software tools. A strong correlation between measured and simulation results were observed. A more detailed model was also created for the Engineering and Applied Science (EAS) Local Area Network. Various performance parameters results were collected and analyzed. Based on simulated results, predictions were made in regards to the scalability and optimization of this network in light of expected future requirements.