Electrical & Computer Engineering     |     Carnegie Mellon

Wednesday, November 5, 12:00-1:00 p.m. HH-1112

 

Phillip Stanley-Marbell
Carnegie Mellon University

Adaptation Methods and Analysis Metrics for Smart Matter

Advances in VLSI technology, and the possibilities offered by non-silicon platforms, are engendering research into the use of large numbers of computational, sensing and actuation devices in what would otherwise be considered mundane non-computational materials. Many challenges exist to the realization of such systems, the least of which might be their actual physical fabrication. The programming, adaptation to failure, and analysis metrics for such platforms are particularly challenging.

This talk overviews our research efforts in this area. A new technique for adapting large numbers of failure prone computational and communication substrates to failure is presented, along with new metrics for characterizing the behavior of failure-prone substrates in the presence of a combination of performance, power consumption and battery lifetime constraints.

Bio
Phillip is a student in the department of ECE at Carnegie Mellon.