ANN model for AlGaN/GaN HEMTs constructed from near-optimal-load large-signal measurements

Microwave neural modeling for silicon FinFET varactors

“…Contrary to the physical and equivalent circuit models, the approach based on artificial neural networks (ANNs) allows obtaining an accurate representation of the device behavior without having to analyze the device structure or the physical processes in it. Among other applications in the field of microwaves [11][12][13][14][15][16], ANNs have been already successfully applied for small-signal and large-signal modeling of different types of high-frequency transistors, such as metal-semiconductor field effect transistor (MESFET), highelectron-mobility transistor (HEMT), metal-oxide-semiconductor field-effect transistor (MOSFET), and heterojunction bipolar transistor (HBT) [17][18][19][20][21][22][23][24][25][26][27][28][29][30] as well as FinFET transistor [31][32][33], which represents an innovative multiple-gate architecture for the downscaling of the complementary metal-oxidesemiconductor technology [34][35][36][37][38]. The present study is aimed at presenting the results achieved by using the ANNs for modeling the scattering (S-) parameters up to 50 GHz for a varactor realized in the advanced FinFET technology, which has already been represented with an equivalent circuit model [9].…”

Combined empirical and look-up table approach for non-quasi-static modelling of GaN HEMTs

“…In the last years, the microwave modelling of gallium nitride (GaN) HEMT has been object of intense studies [1][2][3][4][5][6][7][8][9][10]. This is because such kind of solid state device is attracting a great attention for microwave high power applications, mainly due to the large bandgap of this semiconductor.…”

Untitled

“…In such cases, techniques that allow a wider degree of adaptability are useful. In particular, a new technique that has received increasing attention for the development of PA behavioral models is the neural network (NN) approach [5,6], since model tailoring to the element under study only needs a training procedure based on input-output time-or frequency-domain data sets.…”