Efficient two-dimensional interpolation scheme suitable for nonlinear microwave device modelling

Abstract

The use of device models directly based on measured information is well established in nonlinear microwave circuit analysis, and may provide significant advantages over analytic models relying on parameter extraction techniques. These ‘database’ models always include some sort of interpolation mechanism for overcoming the inherently discrete nature of measured data. This is a delicate aspect, since interpolation may produce unwanted ripples or other ill-conditioning that may considerably degrade the model performance. The authors propose a novel two-dimensional cubic-spline interpolation scheme that guarantees the absence of spurious ripples, and simultaneously provides interpolation of both the measured function and its first derivative.