CAREER: Expecting the Unexpected: A Study of Network Vulnerabilities

This proposal is aimed at studying the resilience of the Internet. The proposer's goal is to answer the question how easy (difficult) it is to break (heal) current networks. Connectivity is a result of the interaction of two mechanisms: the static topology of nodes, links, and routers and the various protocols like routing and DNS that couple these nodes together. The research will be a rigorous evaluation of the complex interaction of the static and dynamic components of networks, the Internet in particular. Will also analyze the recovery behavior of networks, i.e. once a (portion of the) network is attacked or destroyed, how long does it take for the system to return to a state of statistical equilibrium and how far off are the effects of the damage felt. With the PI's prior experience in developing mathematical models for networks, we believe the expected contributions of successful completion of our research are:

Dynamical models of routing mechanisms on the Internet like BGP, OSPF, and IS-IS.

Models for other coupling mechanism on the Internet like DNS and various overlay routing mechanisms,

Modeling and analysis of the couplings between various layers of the Internet protocol architecture

Validation of the various models with simulation/measurement studies; incorporation of the impact of hardware/software implementation details into the models.

Investigation of the interaction of the mechanisms with real topologies.

Development of a mathematical theory of a 'recovery index' for networks.

The proposer believe our efforts will greatly enhance our understanding of the resilience of networks, and both the research as well as operational communities will benefit with the insights gained. The research will be carried out together with a PhD student directly funded by the grant. The PI and the PhD student will regularly publish and give talks on results from the project. Additionally, money has been budgeted for summer internships for undergraduates that will participate in aspects of the project. The New York Academy of Sciences runs a summer program called 'Science Research Training Program', which is an 8 week research internship for high school students. The PI is in touch with the program and high school students will be mentored as part of the project under the program. Columbia University also runs a successful outreach program for Bronx area high schools, offering summer internships for minority and underrepresented groups which we plan to participate in. The PI propose to conduct the simulation and test bed evaluation component of the project with the aid of interested high school students and undergraduates.

The PI teaches a graduate level course on Modeling and Performance Evaluation and a senior level course on Computer Networks on a regular basis. The analytical techniques and models that will emerge as part of the project will be incorporated into the graduate course. The PI has introduced a laboratory component into the undergraduate networks course, and the testbed that is developed as part of the project will be utilized to design novel and interesting lab exercises for the students.