US-Germany Collaboration: Satisfiability-Based Test Generation for Small-Delay Defects in Nanoscale Integrated Circuits

Duke University, Durham NC

Investigators

Abstract

The goal of this collaborative project between Duke University and University of Bremen in Germany is to develop new techniques for detecting small-delay defects (SDDs) in nanoscale integrated circuits (ICs). The PI?s group at Duke University has developed a gate-delay-defect probability measure to model delay variations for nanometer technologies and test-selection techniques to reduce pattern count. The University of Bremen team has carried out extensive research on satisfiability (SAT) and its role in test generation. The proposed collaboration is based on the premise that a probabilistic and pseudo-Boolean SAT framework can be developed for test generation to target SDDs, and the collaborative effort will lead to better-quality solutions than the individual efforts of the two groups. The outcome of this seed project will be a follow-up grant proposal to NSF in the US and DFG in Germany. This project will have a lasting and significant influence on the testing of next-generation nanometer ICs. Theoretical advances in SAT will be driven by and applied to the difficult practical problem of screening SDDs. Lower defect escapes will lead to higher shipped-product quality and lower test cost. Benefits for society include: reliable electronic systems for healthcare; zero-defect ICs for safety-critical systems; cheaper and dependable computing platforms for servers, weather forecasting, climate-change prediction, financial-transaction processing, etc. The innovations resulting from this research will be used to enhance existing courses and develop new courses at both universities. US researchers will benefit from cross-cultural interactions and learn to collaborate with foreign researchers in today?s global workplace.