A Method for Linking Process-Level Variability to System Performances

Tomohiro FUJITA  Hidetoshi ONODERA  

IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences   Vol.E83-A   No.12   pp.2592-2599
Publication Date: 2000/12/25
Online ISSN: 
Print ISSN: 0916-8508
Type of Manuscript: Special Section PAPER (Special Section on VLSI Design and CAD Algorithms)
Category: Simulation
analog VLSI circuit,  statistical analysis,  hierarchical analysis,  response surface model,  intermediate mode,  

Full Text: PDF>>
Buy this Article

In this paper we present a case study of a hierarchical statistical analysis. The method which we use here bridges the statistical information between process-level and system-level, and enables us to know the effect of the process variation on the system performance. We use two modeling techniques--intermediate model and response surface model--in order to link the statistical information between adjacent design levels. We show an experiment of the hierarchical statistical analysis applied to a Phase Locked Loop (PLL) circuit, and indicate that the hierarchical statistical analysis is practical with respect to both accuracy and simulation cost. Following three applications are also presented in order to show advantage of this linking method; these are Monte Carlo analysis, worst-case analysis, and sensitive analysis. The results of the Monte Carlo and the worst-case analysis indicate that this method is realistic statistical one. The result of the sensitive analysis enables us to evaluate the effect of process variation at the system level. Also, we can derive constraints on the process variation from a performance requirement.