Efficient Frequency-Domain Uncertainty Quantification Using Parameterized Model Order Reduction

Abstract

A parameterized model order reduction technique is investigated for the efficient frequency-domain uncertainty quantification of circuits obtained by the Partial Element Equiv-alent Circuit method. The parameterized model order reduction technique is coupled with a standard M C analysis and it is able to provide accurate uncertainty quantification results at a significantly reduced computational cost. Choosing the order of the parameterized model order reduction model is an important step depending on the detail of statistical information needed from the uncertainty quantification process. A practical approach is used for order selection. Numerical results for correlated random variables have validated the efficiency and accuracy of the proposed uncertainty quantification method.