Abstract

Cybersecurity research demands extensive experimentation to be validated. This experimentation is inherently risky: it may involve any combination of live malicious code, disruptive actions and connectivity to the active Internet. Previously, risky experiments were heavily contained to reduce danger to the experimental infrastructure and the Internet. We argue that such an approach unnecessarily stifles research. In this paper we explore a collaborative two-tiered risky experiment management model. It incorporates input from both experimenters and testbed operators; each party specifies constraints on their component’s behavior. Experiment constraints limit behavior in ways that do not affect an experiment's validity, thus meeting usability goals. Testbed constraints build on experiment constraints to meet required safety goals. When combined, experiment and testbed constraints ensure that experiments are both useful to researchers and safe for the testbed and the Internet.

Metadata

publication

Invited paper at the Sarnoff Symposium, 2008

year

2008

publication date

2008/4

authors

John Wroclawski, Jelena Mirkovic, Ted Faber, Stephen Schwab

link

https://deter-project.org/sites/default/files/files/Sarnoff-T1T2.pdf

resource_link

https://deter-project.org/sites/default/files/files/Sarnoff-T1T2.pdf

journal

Invited paper at the Sarnoff Symposium