Academic Projects
Below are descriptions of some of the academic projects that I've worked on, in reverse-chronological order.
Sharing Big Buck Bunny in XIA
For this project, I, along with fellow student, Ben Bramble, worked with a new internet architecture (XIA) to create a video streaming application.
Abstract: The Internet has evolved rapidly over the last twenty years, flowing away from early Telnet designs to a more content-oriented nature. Researchers developed XIA to redesign the underpinnings of the Internet to accommodate not only today's Internet traffic, but allow future innovators to add principals as technology changes. Within this architecture we sought to create a video streaming application using the foundation that XIA provides. We present a novel approach to using DAGs in order to facilitate network-level failover for content chunk delivery. We capitalize on the connectionless nature of XIA's chunking protocol by creating content servers that simply host content without any need for connection-handling. We also use a centralized directory to foster the lookup of CIDs based on video name. Our streaming video application is publicly available and should help guide future XIA application developers.
Honeypots in the Cloud
For this project, I, along with fellow students, Rebecca Lam, Shishir Prasad, Sivasubramanian Ramasubramanian, and Josh Slauson, worked on a project to set up honeypots in Amazon's EC2 in order to profile the kinds of attacks those servers sustain on a daily basis.
Abstract: Honeypots are systems used to trap, monitor, and identify erroneous requests within a network. For this project we conducted a study using honeypots within various cloud computing platforms ( such as Amazon EC2, Windows Azure etc.) with the objective of learning more about what kind of packets they receive. We used various honeypots such as Dionaea, Kippo, and Amun on our cloud instances and gathered data about where attacks came from, what kinds of attacks were made, and differences among cloud instances. We discovered that most attack traffic comes from the US and China and that most attacks are on SSH and HTTP. We also found that for the most part, attack traffic among the clouds was quite similar. We also identified Dionaea and Kippo as the honeypots which are most effective in the cloud setting.