Microwave FET transistor noise modeling using neural networks

Noise modeling of a microwave FET (MESFET) by using a multilayer neural network approach is presented in this paper. We extracted four noise parameters in terms of conventional small-signal intrinsic equivalent circuit elements and three equivalent temperatures. A four layer neural network was used for successful representation. The results obtained by using the neural network are compared with the measured data available in manufacturers´ catalogues.