Unified analytical model of HEMTs for analogue and digital applications

Physical parameter‐based data‐driven modeling of small signal parameters of a metal‐semiconductor field‐effect transistor

“…The modeling of Equation ( 9) and the benchmark is shown in Table 4. A comparison between the Bagging algorithm, Equation (9), and the device simulator result has been achieved by taking the parameters from case 4 in Table 5 and are depicted in Figure 14.…”

“…Modeling of small signal and noise parameters have been achieved either the use of artificial intelligence tools or novel optimization methods. [7][8][9][10] Some other relevant works that uses artificial intelligence algorithms for the modeling of black-box models are: multi-layer perceptron (MLP), support vector regression machines (SVRMs), 7,11 and probability-based neural networks such as the generalized regression neural network (GRNN). 11,12 Black-box-based models can be used for providing a computationally efficient surrogate model for solving the feasible design target space of LNA design optimization problems.…”

“…Small signal and noise parameters of commercially available low‐noise amplifier (LNA) transistors are modeled throughout their operational domain, drain‐source voltage ( V ds ), drain‐source current ( I ds ), and frequency ( f ), independently of their physics‐based parameters. Modeling of small signal and noise parameters have been achieved either the use of artificial intelligence tools or novel optimization methods 7‐10 . Some other relevant works that uses artificial intelligence algorithms for the modeling of black‐box models are: multi‐layer perceptron (MLP), support vector regression machines (SVRMs), 7,11 and probability‐based neural networks such as the generalized regression neural network (GRNN) 11,12 .…”

Signal-noise support vector model of a microwave transistor

“…Such a work is recently published [2] for I-V characteristics of HEMT incorporating the additional second order effects for wider range of operation region including subthreshold, linear, and saturation. In [3], parallel numerical methods are suggested to solve systems of equations generated by the discretization of partial differential equations for the device simulators, which are the most time-consuming part of the simulation process.…”

Full flexible performance characterization of a feedback applied transistor with LNA applications

“…Therefore, the most significant ingredient of an LNA design optimization is the feasible design target space (FDTS), which covers the full potential performance of the transistor in question within its operation ( V DS , I DS , f ) domain. The FDTS must be calculated through the following two stages: (a) the signal and noise parameters of the transistor are modelled throughout its operation domain ( V DS , I DS , f ) using either artificial intelligence tools or novel optimization methods such as those adopted in ; (b) in the second stage, the transistor's highly nonlinear small‐signal performance equations are solved either analytically or numerically with respect to the source ( Z S = R S + jX S ) and the load ( Z L = R L + jX L ) terminations for an operation condition ( V DS , I DS , f ) (Figure ). Within this framework first, a rigorous analytical formulation of the gain G Tmin (f) ≤ G T (f) ≤ G Tmax (f) in terms of the noise figure Freq ≥ Fmin and the input voltage standing wave ratio Vinreq ≥ 1 for an unconditionally/conditionally stable microwave transistor was completed with the { Z S ( f ) , Z L ( f )} terminations in the z‐ and S‐domain, in Güneş et al , respectively.…”