Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Generation of random parameters of behavioral models

: Lange, André; Haase, Joachim; Sohrmann, Christoph; Jancke, Roland

Breitenecker, F. ; Vienna University of Technology:
MATHMOD 2012, 7th Vienna Conference on Mathematical Modelling. CD-ROM : Full Paper Preprint Volume; February 15 - 17, 2012
Vienna: ARGESIM, 2012 (ARGESIM report S38)
5 S.
Conference on Mathematical Modelling (MATHMOD) <7, 2012, Vienna>
Fraunhofer IIS, Institutsteil Entwurfsautomatisierung (EAS) ()
circuit model; statistical analysis; random number generators; Monte Carlo simulation

Technology progress in IC manufacturing is characterized by decreasing the minimum feature sizes used in the fabrication process. With scaling also the impact of the variations increases. The process variations can be divided into inter-die and intra-die variations respectively. It is now interesting in the design process to determine how the known randomness of process parameters influences the randomness of performance parameters of a system such as delay times and energy consumption. It will be shown how randomness of performance parameters as well variations of parameters of behavioural models can be described regarding inter-die and intra-die variations. The approach bases on marginal distributions of the parameters and their Spearman's rank correlation. Non-normal parameter distributions are included. The approach described is a step towards hierarchical statistical modeling of electronic circuits.