A Gaussian Curvature based method for optimizing measurement speed in RF device ANN model

This study reappraises the optimum measurement pattern relying on Load-pull measurement of RF power transistor. The number of load impedance points required to be measured is optimized based on Gaussian curvature. To validate this modeling approach, load impedance points are reduced by 60% on a 0.25 um GaN HEMT device, yet an average prediction loss of -39.17 dB can be achieved for the generation of the ANN model on a range of input powers.