Variation-Aware and Data-Driven Algorithms and Tools for Enabling Reliable Nanascale Design
Event details
| Date | 03.02.2017 |
| Hour | 14:15 |
| Speaker | Dr Zheng Zhang, Massachusetts Institute of Technology USA |
| Location |
ME D2 1124
|
| Category | Conferences - Seminars |
Abstract
The reliability of many engineering systems (such as nanoscale integrated systems, Internet of Things) can be significantly degraded by unavoidable uncertainties (e.g., fabrication process variations, uncertain external environment, and sensor noise). In order to improve circuit and system reliability, variability-aware design automation algorithms and tools should be developed to characterize, estimate and control uncertainties at different levels of design hierarchy. These algorithms and tools are generally data-intensive: a huge amount of data should be generated by non-trivial simulation or measurement, leading to the notorious curse of dimensionality (i.e., the cost grows extremely fast as the number of uncertain parameters increases). This talk presents some fast non-Monte-Carlo techniques to estimate the uncertainties in nanoscale integrated circuits, microelectromechanical systems (MEMS) and silicon photonics. A non-sampling stochastic simulator will be presented to estimate the uncertainties in DC, transient and periodic steady-state analysis. Advanced algorithms will be introduced to efficiently handle many process variations. These techniques can be used as the backbone of a variation-aware design automation flow, and they can accelerate a lot of reliability-related tasks (e.g., yield optimization, chip testing, process modeling, and fault detection). On the IC, MEMS and photonic examples that we have tested, our approaches are significantly faster than Monte Carlo (by 100X to 1000X) and recent spectral methods (by dozens of times).
Bio:
Zheng Zhang is a postdoc associate with MIT, where he received his Ph.D. degree in Electrical Engineering and Computer Science in 2015. He is interested in high-dimensional uncertainty analysis and data inference for diverse engineering problems, including nanoscale devices and systems, hybrid systems (e.g., power systems and robots) and MRI. Dr. Zhang received the 2016 ACM Outstanding PhD Dissertation Award in Electronic Design Automation, the 2015 Doctoral Dissertation Seminar Award from the Microsystems Technology Laboratory of MIT, and the 2014 Best Paper Award from IEEE Transactions on CAD of Integrated Circuits and Systems. He is a TPC member of Design Automation Conference (DAC) and International Conference on Computer-Aided Design (ICCAD).
The reliability of many engineering systems (such as nanoscale integrated systems, Internet of Things) can be significantly degraded by unavoidable uncertainties (e.g., fabrication process variations, uncertain external environment, and sensor noise). In order to improve circuit and system reliability, variability-aware design automation algorithms and tools should be developed to characterize, estimate and control uncertainties at different levels of design hierarchy. These algorithms and tools are generally data-intensive: a huge amount of data should be generated by non-trivial simulation or measurement, leading to the notorious curse of dimensionality (i.e., the cost grows extremely fast as the number of uncertain parameters increases). This talk presents some fast non-Monte-Carlo techniques to estimate the uncertainties in nanoscale integrated circuits, microelectromechanical systems (MEMS) and silicon photonics. A non-sampling stochastic simulator will be presented to estimate the uncertainties in DC, transient and periodic steady-state analysis. Advanced algorithms will be introduced to efficiently handle many process variations. These techniques can be used as the backbone of a variation-aware design automation flow, and they can accelerate a lot of reliability-related tasks (e.g., yield optimization, chip testing, process modeling, and fault detection). On the IC, MEMS and photonic examples that we have tested, our approaches are significantly faster than Monte Carlo (by 100X to 1000X) and recent spectral methods (by dozens of times).
Bio:
Zheng Zhang is a postdoc associate with MIT, where he received his Ph.D. degree in Electrical Engineering and Computer Science in 2015. He is interested in high-dimensional uncertainty analysis and data inference for diverse engineering problems, including nanoscale devices and systems, hybrid systems (e.g., power systems and robots) and MRI. Dr. Zhang received the 2016 ACM Outstanding PhD Dissertation Award in Electronic Design Automation, the 2015 Doctoral Dissertation Seminar Award from the Microsystems Technology Laboratory of MIT, and the 2014 Best Paper Award from IEEE Transactions on CAD of Integrated Circuits and Systems. He is a TPC member of Design Automation Conference (DAC) and International Conference on Computer-Aided Design (ICCAD).
Practical information
- Informed public
- Free
Organizer
- Dean's Office, School of Engineering
Contact
- sylvie Moreau