Research

Most research on the structure of the Internet examines properties of the physical network: how autonomous systems are connected, the resilience of the network to lost connections, and so forth. I am concerned more with the dynamic, virtual networks superimposed on this structure by the ways people actually use the Internet. I study the structure of these virtual networks, the ways in which they interact with the physical network, and the behaviors these structures represent.

I am particularly interested in using statistical properties of these virtual networks to classify different types of network application and identify patterns of behavior that indicate malicious activity.

The data sets I work with in this research are huge (over 30 GB a day), and this has helped me to gain substantial experience in developing and optimizing code for working with large-scale graph data. Some of the networks I analyze contain on the order of 30 million nodes and 300 million edges, which prevents a 32-bit operating system from storing even a sparse representation of their connectivity matrices in main memory.

I am doing this work at the Advanced Network Management Laboratory with the generous advice and support of Steven Wallace, Filippo Menczer, Alessandro Vespignani, Katy Börner, Minaxi Gupta, and Kay Connelly.

There are a lot of people at Indiana University pursuing interesting research in network analysis! I regularly attend Networks and Agents Network meetings every Tuesday and try to maintain close contact with others in the Complex Systems and Networks group.