Seminar: Network Data Analytics in Operational Networks
February 19, 2018 at 11 am in 3107 Patrick F. Taylor Hall
Speaker: Faraz Ahmed, Michigan State University
Large operational networks, such as Cellular Networks, Content Delivery Networks (CDNs), and Online Social Networks (OSNs), produce data that can be analyzed to understand and improve network performance. My research efforts have been aimed towards providing network operators with tools, techniques and actionable insights to cope with the rapidly evolving networks by measuring network performance and capturing new trends through modeling and analysis of network data. In this talk, I will present our approaches for measuring, modeling, and analyzing performance of cellular networks and content delivery networks.
First, I will present our work on detection and localization of performance degradation in cellular networks. Cellular network users can experience performance degradation that can occur due to issues outside the cellular network, it may be due to the user device itself (such as device OS, application software) or due to the content provider (such as application servers, datacenter network). We focus on the detection and localization of End-to-End performance degradation across four administrative domains: cellular network, content providers, device manufacturers, and application developers. I will also discuss our approach to assess the user impact of issues that occur within the cellular network. Second, I will present my research on measurement, modeling, and analysis of client perceived network performance in operational CDNs. Two major considerations for CDNs are cost and performance dynamics of delivering content to end users. CDNs maintain multiple transit routes from content delivery servers to eyeball ISP networks which provide Internet connectivity to end users. I will present our approaches towards optimal transit route selection for CDNs. Finally, I will briefly discuss our work on privacy preserving social network data publishing
Faraz Ahmed is a Ph.D. candidate in the Department of Computer Science and Engineering at Michigan State University. His research interests lie in measurement, modeling, and analysis of large scale network data. Specifically, he has worked on research problems that involve network performance data analysis of Cellular Networks and Content Delivery Networks. He is a recipient of the 2017 Fitch H. Beach Award for Outstanding Graduate Research and the 2017 Outstanding Graduate Teaching Award, awarded by the College of Engineering and the Department of Computer Science & Engineering at Michigan State University.