JOINT MATHEMATICS COLLOQUIUMUNIVERSITY OF IDAHOWASHINGTON STATE UNIVERSITY |
|---|
|
Abstract |
|---|
| To engage in the practical design and
construction of electronic components and systems, one must mitigate
the presence of uncertainty in the design and manufacturing process
which leads to material and structural variability in the realized
system. To improve production yields of widely multi-scale structures
under conditions of uncertainty in a high volume manufacturing (HVM)
environment, the designer resorts to electrical design automation (EDA)
tools in the search for a robust solution space. In this context, the
accuracy and efficiency of the underlying models and methods of EDA are
fundamentally critical to the characterization efforts of the
components and system, and in providing indispensable intuition and
guidance throughout the design process. To enable an efficient
stochastic-based design optimization methodology for electrical
components and systems, which exhibit randomness in their material and
geometric characteristics, we infuse the stochastically collocated
reduced-order state-space electromagnetic model, into one of the most
extensively used methods for electromagnetic modeling and simulation,
namely, the method of finite-difference time-domain (FDTD). To this
end, we develop the stochastic electromagnetic macro-model in FDTD by
formulating an abstraction layer that encapsulates the fine features of
the multi-scale structure, where uncertainty is most often present.
|