In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
will defend his dissertation
NetForager: Geographically-Distributed Web Application Traffic Generator
AbstractUsers connect to web applications for various content and services. These connections create the largest traffic on the internet. Data scientists can better understand network usage by analyzing internet-based network traffic. Network traffic analysis requires an applicable dataset. While there are a few public datasets, most researchers prefer to collect private traffic traces. This approach deprives those in academia the opportunity to build on top of the collected research and further advance the network traffic analysis field. For other data scientists to use any published network traffic details, the data must not contain sensitive user information. Also, to use the gathered traffic results in other network analysis efforts, scientists need generic data with repeatable methods of collection that is portable between various platforms. Therefore, data scientists need a framework that can easily and thoroughly capture network traffic details, allowing the user to configure and extend the framework, to capture data anonymously for sharing purposes, to isolate targeted data and to automatically label the retrieved data. To answer this need, I have developed a data collection framework called NetForager that will help data scientists thoroughly investigate network traffic. NetForager provides an easy way to collect network traffic data in a reproducible and repeatable way with minimal human intervention. After analyzing web application traffic using NetForager, my results revealed interesting patterns, which could be used to build an algorithm to uncover more web application data. I collected a data subset from web applications. Then, I applied foundational network analysis to the collected dataset. This analysis resulted in the discovery of unique web application behaviors many of which pose considerable challenge for if analysis were to be done on data collected through the existing data collection methods. As a result of the NetForager development, I was able to create foundational building blocks in analysis of web application network traffic that the data scientists can now use. The foundational building blocks can be used by the data scientists to further develop more algorithms for the observation of other network usage trends in a repeatable and portable manner advancing network science and engineering research in the long run.
Date: Thursday, November 14, 2019
Time: 11:30 AM - 12:30 PM
Place: CoT 204
Advisor: Dr. Deniz Gurkan
Faculty, students, and the general public are invited.