A behavioral modeling approach to nonlinear model-order reduction for RF/microwave ICs and systems

This paper considers an approach to nonlinear model-order reduction for RF/microwave integrated circuits (ICs) from the perspective of "black-box" behavioral modeling. We present a systematic methodology for creating behavioral models using techniques developed from concepts in system identification, nonlinear dynamics, computational geometry, and information theory. Highly complex subsystems can be represented by relatively straightforward input-output relationships involving the observed and identified dynamical variables. Model order is thus significantly reduced compared with the device-level representation. We illustrate the technique by creating a cascadeable transportable model of a wide-band microwave IC amplifier that accurately predicts the dc, large-signal, harmonic and intermodulation distortion, and small-signal (S-parameter) behavior.