In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy
will defend his dissertation proposal
Realistic Traffic Generation Through Application-Agnostic Learning
Research and testing in networking requires experiments to be performed with real or representative network traffic. Privacy policies often prevent experimentation with production traffic or captures by third parties, which includes most academic researchers. Presently, most experimenters rely on synthetic traffic generators that send packets at fixed rates, or at rates based on statistical distributions; others replay captured packets traces, which often have limited durations. To solve this problem, we propose to create ‘traffic models’ for the patterns of network traffic seen in production networks by using machine learning (ML) algorithms to model applications’ network behaviors from traffic traces, after removing all protocol-specific reactions to network impairments. The ML system, running on a third-party server, would receive real production traffic and the output would be traffic models that can be taken to a totally different research environment for re-generation of the traffic, without the researcher gaining any knowledge about the input training set. The outcome and methods derived from this research can be a game-changer in the way experimentation on large scale networks (datacenter, cloud, enterprise and IOT) is done especially in academia.
Date: Friday, December 7, 2018
Time: 1:15 PM
Place: T2 200
Advisors: Dr. Deniz Gurkan
Faculty, students, and the general public are invited.