“…If much more neurons are needed to accurately model the nonlinearities, then a multilayer neural network can also be used [11] to reduce the number of neurons.…”
Section: Artificial Neural Networkmentioning
confidence: 99%
“…In [8,9], neural networks are used to design a microstrip corporate feed and also to optimize interconnects in high-speed VLSI circuits. ANNs have also been used to model the high nonlinearities of transistors obtained by measurements or empirical models [10,11] The extended FDTD approach suggests to include nonlinear devices in the FDTD grid as in [12][13][14]. This method leads to an equivalent circuit that characterizes the wave-device interactions, which gives a good approximation of a global modeling simulation providing that the transistor model comes from a semiconductor simulation.…”
We propose a first-order global modeling approach of Monolithic Microwave Integrated Circuits (MMIC) by modeling the active device with a neural network based on a full hydrodynamic model. This neural network describes the nonlinearities of the equivalent circuit parameters of an MESFET implemented in an extended Finite Difference Time Domain mesh to predict large-signal behaviors of the circuits. We successfully represented the transistor characteristics with a one-hidden-layer neural network, whose inputs are the gate voltage V gs and the drain voltage V ds . The trained neural network shows excellent accuracy and dramatically reduces the computational time in comparison with the hydrodynamic model. Small-signal simulation is performed and validated by comparison with HP-Libra. Then large-signal behaviors are obtained, which demonstrates the successful use of the artificial neural network.
“…If much more neurons are needed to accurately model the nonlinearities, then a multilayer neural network can also be used [11] to reduce the number of neurons.…”
Section: Artificial Neural Networkmentioning
confidence: 99%
“…In [8,9], neural networks are used to design a microstrip corporate feed and also to optimize interconnects in high-speed VLSI circuits. ANNs have also been used to model the high nonlinearities of transistors obtained by measurements or empirical models [10,11] The extended FDTD approach suggests to include nonlinear devices in the FDTD grid as in [12][13][14]. This method leads to an equivalent circuit that characterizes the wave-device interactions, which gives a good approximation of a global modeling simulation providing that the transistor model comes from a semiconductor simulation.…”
We propose a first-order global modeling approach of Monolithic Microwave Integrated Circuits (MMIC) by modeling the active device with a neural network based on a full hydrodynamic model. This neural network describes the nonlinearities of the equivalent circuit parameters of an MESFET implemented in an extended Finite Difference Time Domain mesh to predict large-signal behaviors of the circuits. We successfully represented the transistor characteristics with a one-hidden-layer neural network, whose inputs are the gate voltage V gs and the drain voltage V ds . The trained neural network shows excellent accuracy and dramatically reduces the computational time in comparison with the hydrodynamic model. Small-signal simulation is performed and validated by comparison with HP-Libra. Then large-signal behaviors are obtained, which demonstrates the successful use of the artificial neural network.
“…Several modeling approaches based on artificial neural networks and belonging to the second category of solutions have been presented in the literature [9][10][11]. Neural networks have the ability to simulate nonlinear relations with high accuracy.…”
Abstract-In this paper, a new technique is proposed for field effect transistor (FET) small-signal modeling using neural networks. This technique is based on the combination of the Mel frequency cepstral coefficients (MFCCs) and discrete sine transform (DST) of the inputs to the neural networks. The input data sets to traditional neural systems for FET small-signal modeling are the scattering parameters and corresponding frequencies in a certain band, and the outputs are the circuit elements. In the proposed approach, these data sets are considered as forming random signals. The MFCCs of the random signals are used to generate a small number of features characterizing the signals. In addition, other MFCCs vectors are calculated from the DST of the random signals and appended to the MFCCs vectors calculated from the signals. The new feature vectors are used to train the neural networks. The objective of using these new vectors is to characterize the random input sequences with much more features to be robust against measurement errors. There are two benefits for this approach: a reduction in the number of neural networks inputs and hence a faster convergence of the neural training algorithm and robustness against measurement errors in the testing phase. Experimental results show that the proposed technique is less sensitive to measurement errors than using the actual measured scattering parameters.
“…In [24,25], Sirakawa et al have proposed a multiplayer perceptron structure for large signal model of HEMT. For the proposed ANNbased small-signal modeling approach in this paper, a sub-ANN (SANN) for each intrinsic element is adopted, the modeling idea is illustrated in Fig.…”
Abstract-An ANN-based small-signal equivalent circuit model for 130 nm MOSFET device is proposed in this paper. The proposed model combines the conventional small-signal equivalent circuit model and artificial neural networks (ANNs) to achieve higher accuracy. Good agreement is obtained between proposed model and measured results confirming the validity and effectiveness of proposed model.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.