Efficient Uncertainty Quantification of FDTD based Microwave Circuit Models with Multiple Design Parameters

The polynomial chaos expansion (PCE) method has emerged as a promising uncertainty quantification technique compared to the commonly used yet computationally inefficient Monte Carlo method. However, PCE-based methods generally suffer from a "curse of dimensionality", where the computational cost increases rapidly with the number of random variables included in the analysis. This paper applies an orthogonal matching pursuit algorithm to mitigate the computational cost of PCE and facilitate the uncertainty analysis in finite-difference time-domain (FDTD) models of microwave circuits. The performance is demonstrated by modeling a cascaded stub filter with 13 geometric and material parameters, where a considerable computational advantage is achieved.