Research Activities using Opnet

Dr. Yi Du developed a general technique to detect and control congestion status by using special probe and feedback packets.

Congestion detection consists of having the source send special probes during a specified period of time and then await the return of a feedback packet from the destination, which conveys the result of the analysis performed by the destination upon reception of the probe packets.

Yi Du's technique is based on the notion of detection interval, which represents the time to complete a basic congestion detecting interaction between a packet source and destination.

Opnet is used to simulate high-speed packet networks in order to evaluate the timeliness of the information gathered by this technique, i.e., after sending a probe and receiving feedback, how much has the traffic network changed with respect to the information gathered.

2. Monitoring and control of network traffic

The objective of this project is to develop strategies for automatic routing after observing and analyzing traffic behavior on the web. The research involves using active system techniques to identify potential and real transaction requests within specific systems.

Through statistical learning and active data mining techniques we seek to detect patterns in the requests for transactions to state association rules and sequence of frequent patterns.

The knowledge of preferred transactions will allow the prediction of routes and destinations requests to adjust server behavior to accommodate in advance the network status expected. The result will be an enhancement of performance by reducing traffic bottlenecks and optimizing bandwidth.

Opnet Modeler allows the creation of a virtual networking test bed to implement the strategies and measure their performance.

3. Comparison of theoretical and experimental queue models.

In this project Opnet is used to simulate customer-processor scenarios involving different types of queuing to analyze their behavior and compare them with respect to the one predicted by the Queuing Theory.

Teaching Activities using Opnet

CS444 is an advanced undergrad course taught by Dr. G. M. Masson, that covers intrasystem communication issues like layered network architectures; the OSI model; signals, bandwidth, modulation, data rates, modems, multiplexing, error detection/correction; switching; queuing models, circuit switching, packet switching; performance analysis of protocols, local area networks; and congestion control.

Students practice these concepts through several ready-made Opnet labs. Also, students learn Opnet Modeler fundamentals by modeling the link layer of a networking system.

2. Authoring of lab projects and tutorials.

• Automatic Request Protocols - Jorge Vasconcelos
• Queue models - Chris Kyvernitis
• Error detection - Chris Kyvernitis
• Aloha Protocol - Natalia Escribano
• Opnet Fundamentals Guide - Jorge Vasconcelos
• Basic Modeling with Opnet - Zhijuan Mao