KEYNOTE: Mitigating Emergent Computation: the need for new approaches in systems engineering
Presented as part of the 2021 HCSS conference.
ABSTRACT
Modern computing systems demonstrate strong propensity for unintended, emergent computations and the related unintended, emergent programming models that enable or amplify cyber-attacks. Computing mechanisms built for a particular purpose and with particular intended models of execution in mind prove to be capable of executing unintended computing tasks outside of their original specification and their designers and programmers' mental models. The Spectre family of attacks on CPU microarchitectures is one recent example, with more examples provided by operating systems kernels and standard runtime environments.
Today, we start examining systems for signs of emergent behavior, with methods such as fuzz-testing, only after they are fully built. However, recent research strongly suggests that a system's exploitability and propensity for emergent execution arise, and can also therefore be mitigated, already at the design stage.
The talk will discuss the challenges in systems engineering, formal methods, as well as in program analysis and modeling to construct systems of systems in which emergent computation is mitigated.
Dr. Sergey Bratus joined DARPA as a program manager in January 2018. His research interests include computer security and intrusion analysis.
Dr. Bratus joins DARPA from Dartmouth College, where he has worked since 2002, initially as a post-doctoral research associate and most recently as a research associate professor in the Computer Science department. He has made significant technical contributions in the area of language-theoretic security, which aims to eliminate broad classes of input-driven exploitable vulnerabilities. Dr. Bratus began his professional career at BBN Technologies, where he worked on natural language processing.
Dr. Bratus holds a Doctor of Philosophy degree in mathematics from Northeastern University. He has published over 100 technical papers, and his research has received several awards, most notably the 2012 BlackHat Pwnie award for Most Innovative Research. Dr. Bratus also earned best paper awards at the WOOT 2011, IJSSE 2010, and TRUST 2008 conferences.