In this thesis, a fully statistical multivariate modeling approach is introduced to describe variability in integrated circuit design. The approach supports Gaussian as well as almost arbitrary non-Gaussian distributions and captures correlations. Thus, it allows analyzing the variability of multiple performance parameters jointly and accurately. Application scenarios at different levels of abstraction in integrated circuit design demonstrate the versatility and accuracy of the presented approach.     «

In this thesis, a fully statistical multivariate modeling approach is introduced to describe variability in integrated circuit design. The approach supports Gaussian as well as almost arbitrary non-Gaussian distributions and captures correlations. Thus, it allows analyzing the variability of multiple performance parameters jointly and accurately. Application scenarios at different levels of abstraction in integrated circuit design demonstrate the versatility and accuracy of the presented approac...     »