This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5). The award supports participation of U.S. graduate and undergraduate students from the PI?s Mixed-Signal Nanometer VLSI Design Lab in a collaboration on the characterizations and design methodologies for integrated circuit (IC) under process variation impact. Device variability tends to spiral out of control as circuit design technology enters the nanometer design regime. Most current research on adaptive design and statistical analysis tools emphasizes that the manufactured chips match the original design specifications. Several tough issues prevent the diminution of variability. For example, how do we model the variability? How do we quantify it? How do we manage the variability that stems from irreducible complexity? Is it possible for us to make use of the variability instead of just controlling it? The proposed joint research intends to provide novel approaches to overcome these challenging obstacles.
The PI and her students will collaborate with Prof. Charlie C.P. Chen from Department of Electrical Engineering, National Taiwan University, and Prof. Guanrong Chen and Prof. Wallace K.S. Tang, both at the Centre for Chaos and Complex Networks (CCCN), Department of Electronic Engineering, City University of Hong Kong. The award support four U.S. students?two undergraduates and two graduate students?each summer over the next three years to conduct the proposed joint research. The unique, open sources on process data at National Taiwan University offers first hand data to support research on the modeling and characterization of design for manufacture. The facilities of City University of Hong Kong provide an exceptional opportunity to combine the expertise of U.S, Taiwan and Hong Kong scientists in a synergistic program to elucidate the property of process variation mechanism and develop design methodologies to utilize the variability.
In the first year students will exploit the idea of statistical performance to model device and gates. The newly proposed models are to be tested on 45 nm process data in cooperation with the National Taiwan University. Expected laboratory tasks include the measurements of the performance distribution and of environmental influences on IC design. Several different statistical models including the finite point model formulated at The City University of Hong Kong may be applied to the experimental measurement data. During the second year, students will investigate new strategies that employ performance oriented parameter variation reduction and provide performance bound estimation. They will test the ability of the proposed models to characterize and verify different types of mixed-signal circuits. In the third year, students are expected to integrate the above research components into a new design flow. A highly nonlinear circuit proposed by The City University of Hong Kong will provide a prototype circuit. Students will redesign the circuit using the new flow, measure and test the circuit performance, and verify the quality of design.
The project will increase understanding of the statistical nature of process variations, the impact on circuit performance, and their potential for recovery following an intelligent design algorithm. In this program, students will develop scientific hypotheses, design individual research projects, and analyze and present their results orally and in written form. They will be exposed to laboratory and field research techniques in a unique environment. The discoveries from the joint research to be undertaken will also inform the development of curriculum and fill a gap in current textbook coverage of process variation and chaotic modeling as a scholarly discipline.
