Techniques for Application -Specific Circuit Customization and Adaption

Farinaz Koushanfar
Rice University

Abstract

I present our ongoing efforts in designing and building customizable and adaptive embedded systems.
For efficiency, energy, and speed reasons, many intensive processing units are increasingly
implemented using custom hardware. The major limitation of conventional customized design is lack
of flexibility and adaptability. The address this limitation, we have developed an N-variant methodology
and low overhead implementation algorithm for application specific integrated circuit (ASIC). N-variant
design is an approach for realizing  multiple variants of the original specification in the same
implementation. Each variant has the same input/output characteristics and interfaces as the original
specification. However, the internal structure or the orders of the internal events are specific to each
variant. Integrating multiple copies of the specification enables flexibility by allowing a degree of freedom
in statically or dynamically selecting the implementation variant before or during the operation. I discuss
how the management of the different variants could enable a range of applications including fault tolerance,
post-silicon optimization, and hardware IP protection and digital rights management. I also briefly present
our ongoing work on IC characterization as the precursor for many customizable and adaptive design
applications.

Bio:

Farinaz Koushanfar is an Assistant Professor in the Departments of Electrical & Computer Engineering
and Computer Science at Rice University, where she is also the Director of Texas Instruments DSP
Leadership University Program. Before joining Rice in 2006, she received her Ph.D. in Electrical
Engineering and Computer Science and her M.A. in Statistics both from UC Berkeley. Her research focus
is in the area of embedded systems. She creates techniques for synthesis and design of embedded
systems with an emphasis on customizable, adaptive, lightweight, and secure devices. Her ongoing
projects are focused on customized and adaptive embedded systems, protection and security of hardware
embedded systems, adaptive energy sources, and applications of emerging technologies. For her research,
she has received a number of awards and honors including Office of Naval Research Young Investigator
Program (YIP) Award, National Science Foundation (NSF) CAREER Award, Young Faculty Award from the
Defense Advanced Research Projects Agency (DARPA), INTEL Open Collaborative Research (OCR)
Fellowship, and a Best Paper Award at Mobicom. In 2008, she was named one of MIT Technology Review's
Young Innovators Under 35 (TR-35).