| Author(s): |
Kevin L. Mills; James J. Filliben; |
| Title: |
AN EFFICIENT SENSITIVITY ANALYSIS METHOD FOR NETWORK SIMULATION MODELS |
| Published: |
December 07, 2010 |
| Abstract: |
Simulation models for data communications networks encompass numerous parameters that can each take on millions of values, presenting experimenters with a vast space of potential parameter combinations. To apply such simulation models experimenters face a difficult challenge: selecting the most effective parameter combinations to explore, given available re-sources. This paper describes an efficient method for sensitivity analysis, which can be used to identify significant parameters influencing model behavior. Subsequently, experimenters can vary combinations of these significant factors in order to exercise a wide range of model behaviors. The paper applies the sensitivity analysis method to identify the most significant parameters influencing the behavior of MesoNet, a 20-parameter network simulator. The method and principles explained in this paper have been used to investigate parameter spaces for simulated networks under a variety of congestion control algorithms |
| Proceedings: |
Proceedings of the 2010 Winter Simulation Conference |
| Location: |
Baltimore, MD |
| Dates: |
December 5-8, 2010 |
| Keywords: |
discrete event simulation; experiment design; sensitivity analysis |
| Research Areas: |
Modeling |
| PDF version: |
 Click here to retrieve PDF version of paper (788KB) |
