Research Experience for Undergraduates – Summer 2011
eXpressive Internet Architecture project

The eXpressive Internet Architecture (XIA) project is developing a new network architecture that will significantly improve the trustworthiness, flexibility, performance, usability and evolvability of the Internet. It is a large project that involves faculty in networking, security, human-computer interfaces, and engineering and public policy. More information can be found on the XIA web site: http://www.cs.cmu.edu/~xia

XIA includes many focused research efforts that can benefit from the participation of undergraduate students. Three such projects are listed below. Please contact the faculty member(s) listed with a CV and transcript if you are interested.

Anonymity and Information Reuse

Attributes of an information source can affect how recipients attend to and interpret content. Information from anonymous sources may be viewed as untrustworthy because users have little or no means for sanctioning poor or inaccurate content. Further, when the potential for information overload is high, the absence of personal identifying information makes sorting through anonymous information exhausting, thereby limiting its reuse. Thus its dissemination might be limited by poor user adoption. On the other hand, because anonymity reduces the risks associated with information sharing such as embarrassment or social sanctions, it can facilitate the transmission of controversial, critical, or novel information. Accordingly, information arising from anonymous sources has the potential to be highly influential.

We are looking for two undergraduate students to work with us on investigating the relationship between anonymous information and its reuse. The purpose is to gain a better understanding of the barriers that facilitate or limit the transmission of anonymous information. Students will be involved in conducting reviews of the academic literature relating to anonymity and information reuse; compiling and summarizing examples of how anonymity affects information use in online environments, and conducting experiments with study participants.

We plan two distinct projects. One project will look at information reuse in online environments, and the other conducting experiments in the lab. For the second project, the student will assist in developing the experimental task, study materials, and running experimental sessions. The students will gain valuable behavioral and technical experience conducting research and running experiments.

Interested students should contact Professor Sara Kiesler (kiesler@cs.cmu.edu ).

Implementing XIA forwarding on Openflow

A team of graduate students working on the core XIA architecture has developed a prototype XIA router and host communication library based on the Click protocol framework on Linux. We are looking for two REU undergraduate students to implement XIA forwarding for the host, content, and service principals on the Openflow platform, an industrial strength network switch. This project will address three key shortcomings of the existing prototype system. First, since Click runs on general purpose hosts, it is not at all representative of how today’s commercial switches and routers forward packets; Openflow is a better approximation. Second, the Openflow implementation should be substantially faster than the Click implementation, allowing more realistic experiments involving larger numbers of flows. Third, GENI is deploying Openflow switches, so the Openflow XIA forwarding engine will allow us to run larger scale XIA experiments over GENI.

In combination, these three improvements will allow us to study the implementation cost and forwarding performance of the XIA forwarding engine in a more realistic context. In turn, this will allow us to research the tradeoffs between XIA’s expressiveness and runtime complexity, both with respect to some of current XIA features (e.g. the use of an address format based on DAGs). In addition, it will enable work on future research questions such as how to improve scalability by adding different types of hierarchy to the addresses for the various principals, or how to efficiently enforce a provider’s policy constraints while processing packets. Students participating in this project will learn about the design tradeoffs in the core of the Internet.

Interested students should contact Professors Peter Steenkiste and Srini Seshan (prs@cs.cmu.edu and srini@cs.cmu.edu ).

Applications Leveraging XIA

Two unique features of XIA are support for multiple communicating principal types and the intrinsic security properties associated with all communication operations. One of the goals of these features is to simplify the development of trustworthy network applications; however, the design and implementation of these two features is also closely related to the way applications will use them. For example, XIA allows applications to express their communication intent through the choice of communicating principal type. A file sharing application might use a content principal to indicate that the communication goal is to fetch content, while a ssh-like application might use a host principal to indicate its desire to contact a specific host. The types of principals that the network supports reflects the range of communication “modes” that applications require. In a similar fashion, XIA relies on communication principals to bootstrap application-specific security properties for each communication operation. Since the design of XIA is closely tied to applications, it is important that we evaluate the impact of XIA on application development.

This third project involves developing a distributed application over XIA. While the choice of application depends on the preferences and background of the students involved, a good example would be a social networking application since it involves multiple communication modes (e.g., exchanging files, contacting/updating users, group communication, and services such as finding friends or online games) and has significant privacy and trust implications. The focus of the effort is on building this application using the key XIA features (multiple principals, intrinsic security) and evaluating the benefits of using XIA for this application development over the existing Internet protocol suite. We plan to expand the scope of this effort in future years.

Interested students should contact Professors Peter Steenkiste and Srini Seshan (prs@cs.cmu.edu and srini@cs.cmu.edu ).