opnet1.bmp (155518 bytes)
opnet2.bmp (8098 bytes)
OPNET Technologies
3400 International Drive, NW
Washington, DC 20008
Tel: 202-364-4700
Fax: 202-364-8554
E-mail: mailto:info@mil3.com
Web: http://www.mil3.com/home.html

OPNET is a registered
trademark of OPNET Technologies
© 2000 OPNET Technologies
University: Duke University
Name of Sponsoring Professor: Kishor S. Trivedi
Department: ECE

1. Performance study of wireless networks

We plan to model and analyze the performance and quality of service (QoS) of wireless networks, with consideration of the operation reliability/survivability issues. This topic has attracted ever-increasing research interest when the industry is becoming more aware of disaster avoidance/recovery capability of their networks. We plan to build up a simulation model for a typical wireless network based on some existing protocols (IEEE 802.11 as an example), and compare the results obtained from simulation with analytical results obtained from analytical models built with stochastic Petri nets.

 2  Performance analysis of RED routers supporting TCP flows

The performance study of TCP connections has drawn much research attention and considerable effort in recent years.  More recently, other queue management mechanisms have been proposed, and among them, Random Early Detection (RED) is the most popular.  The aim of RED is to improve the performance of TCP connections by minimizing the drawbacks of the traditional Drop-tail policy.  Although many variants of RED and its generalized mechanisms, active queue management (AQM) have been proposed, in general, the choice of parameters of RED or AQM has remained more of an art than a science, making the setting of control values at least case dependent. Setting the parameters and analyzing the performance associated with such schemes, which is important for performance improvement, might be a real challenge in an operational router. Simple theoretic guidelines are needed for the effective design of RED scheme by relating its parameters to network parameters such as propagation delay.

 In this research, we propose a mathematical model as a stochastic feedback control system for TCP flows with heterogeneous round trip times, by means of stochastic differential equation theory.  Based on this simple model, we give the performance and stability analyses of RED congestion control modules, and then give an explicit formula to exhibit the relation among the RED parameters as well as the network parameters.  The analytical results are verified by discrete-event simulations.

Publications:

1. Helena Szczerbicka,  Kishor S. Trivedi  and Pawan K. Choudhary, DISCRETE EVENT SIMULATION WITH APPLICATION TO COMPUTER COMMUNICATION SYSTEMS PERFORMANCE, to be published in World Computer Congress, 22-27 August, 2004