Parallel Simulation of Telecommunication Networks

Main objective of this research is to build a high-performance discrete event simulation framework for performance analysis of large-scale broadband telecommunication networks. We plan to use the framework to analyze the performance of protocols for provision of QoS (Quality of Service) and predictive congestion control mechanisms in large-scale telecommunication networks via simulation studies. We specifically intend to study delivery of multimedia through the wired and wireless networks.

In this project, we are developing a high-performance parallel discrete event simulation system which will be capable of exploiting the computational capacity of clustered supercomputing. We aim to achieve this by modifying the OMNeT++ simulation system kernel.

Discrete event simulation is a well established technique for performance analysis of telecommunication networks which are complex and distributed systems involving time varying parameters, inherent nonlinearities and uncertainties. Analytical models can not cope with this level of complexity, especially after the establishment of the failure of Poisson models in packet switched networks. Instead, simulation is now considered a critical tool in developing, testing and evaluating network protocols and architectures. It provides a practical methodology for understanding network system behaviour that is either too complex for mathematical analysis, and/or too expensive to investigate by measurements or prototyping on real network equipment.

Today, due to the emergence of global networks and gigabit speed links, simulation of communication networks is faced with significant challenges:

Such characteristics cannot be captured by simulation of small network models. Because of this, there is an increasing need to simulate large-scale communication networks for extended time periods. Simulation of communication systems at this level of complexity requires large amount of memory and computing power that is susually available only on supercomputers. Additionally, it is widely acknowledged that the capabilities of conventional simulation techniques are not sufficient to address such simulation requirements, and parallel/distributed simulation techniques must be employed.

Currently efforts are being made on development of supercomputing clusters as cost-effective high-performance computational platforms. For the reasons mentioned in the previous paragraph, and increasing availability of supercomputing clusters, we concentrate our efforts on development and analysis of algorithms and traffic models for efficient, realistic simulations of very large-scale communication networks on clustered parallel computer systems.

For further information about this project Ahmet Sekercioglu can be contacted.

[back to Ahmet's home] [back to CTIE home] [back to E&CSE home] [back to Monash home]
This page has been accessed times since 16/January/2001.
Last update by Ahmet Sekercioglu (Date: 2000/12/05 04:19:27 $ $Revision: 1.3 $).