Search

The purpose of this experimental study is to evaluate quantitatively the impact of the temperature on the behavior of various high electron-mobility transistor (HEMT) technologies through the analysis of the DC and RF transconductance. The experimental data are reported for six different HEMT devices, in order to develop a comparative analysis based on various technologies, including gallium arsenide (GaAs) and gallium nitride (GaN) materials, matched and pseudomorphic HEMTs, single- (S-H) and double-heterojunction (D-H) HEMTs, and both virgin and multi-layer devices. The reported findings show that the impact of the ambient temperature on the HEMT behavior strongly depend on the tested technology and operating conditions. As a matter of fact, a higher temperature can lead to increased or degraded transconductance, depending on the device technologies and bias point. In the GaAs-based devices, an operating bias condition at which the DC and RF transconductance are temperature insensitive can be defined, owing to two-opposite temperature-dependent effects counteracting with each other.

In this research, four topologies have been investigated to characterize substrate/buffer loading effect on GaN HEMT on Si. This effect has been simulated using open de-embedded structure’s Z-parameter measurements for GaN HEMT on Si. An optimization technique based on Grey Wolf Optimization (GWO) with Cuckoo Search (CS) has been developed to extract four models’ elements. The simulated data fitting compared with the measured one has been considered the validation of the process. The models are evaluated in terms of rate of convergence to minimum error and accuracy of the results. The results showed good agreement with previous literature.

This paper presents a simple approach to model the self-heating effect in GaN high electron mobility transistors (HEMTs) using a particle swarm neural network and also reports the extraction procedure of the model parameters. The main advantage of the developed method is its simplicity of construction and implementation in computer-aided-design tools. The developed modeling procedure is applied to a packaged GaN HEMT and validated by DC and AC small/large-signal simulations, which showed a very good agreement with the measurements.

In this study, a 4.4 MHz wireless power transfer (WPT) system was designed and implemented using a class-AB Armstrong oscillator that uses a packaged GaN on Si high electron mobility transistor. The oscillator is designed and simulated in advanced design system and then integrated with a magnetically coupled resonant WPT system. The transmitting planar coil of the WPT contains both the main and the feedback coils of the Armstrong oscillator. The whole system was implemented in a printed circuit board and tested. The DC-to-AC conversion efficiency of the simulated Armstrong oscillator is 69%. The maximum measured power transfer efficiency is 40.3% at a 2 cm distance between the transmitting and receiving coils. The input DC power of 37.5 mW provides 15.1 mW of AC power to the load with low distortion, making it suitable for low electromagnetic interference, size, power and cost applications such as biomedical implants.

In this research, four different equivalent circuit models that characterize the substrate/buffer loading effect for GaN HEMT on Si substrate have been investigated. Z-parameter measurements of an open de-embedded structure have been used to characterize this effect in 16×200 μm GaN HEMT on Si. An optimization technique based on hybridization between two meta-heuristic techniques which are Particle Swarm Optimization (PSO) and Grey Wolf Optimization (GWO) has been developed to extract the models' elements. Each of those techniques has been tested separately to evaluate the performance of the proposed technique. The validation process of the extracted elements is done by comparing the simulated and the measured data of the model. The accuracy and the rate of convergence of the extraction results are considered as evaluation criteria. The proposed PSO-GWO technique shows a very good agreement with previous research results, which accordingly validates the applicability of this technique for small- and large-signal modeling of transistors.

This work investigates the applicability of a two-stage Colpitts oscillator based on GaN high electron mobility transistors (HEMT) for wireless power transfer applications. An inductive load is used both as part of the resonance network in the oscillator and as a power transmitting coil. Simulations and experimental results using commercial GaN transistors are provided.

This paper presents three different approaches for designing dc-ac converter based on GaN transistor. The proposed circuits could be used to design the transmitter side of Wireless Power Transfer (WPT) systems. The first two approaches are based on single-stage and double-stage Colpitts Oscillators, and the third one is based on an H-bridge converter. The three circuits have been simulated using LTSpice software, then implemented on PCB boards. The whole WPT system has been implemented, cased, and tested.

In this paper, a new modeling technique is proposed for extracting small-signal lumped-element equivalent-circuit models for microwave transistors. The proposed procedure is based on using an optimization approach that is improved by targeting a quasi-static behavior as additional objective function rather than only minimizing the error between the simulated and measured scattering parameters. The validity of the developed modeling methodology is successfully demonstrated by considering a 0.25x1000 $\mu \text{m}^{2}$ gallium nitride (GaN) high-electron-mobility transistor (HEMT) as a case study.

Forecasting of Electric Peak load is very essential routine work for safe operation of power systems. In this paper, three different forecasting approaches are demonstrated. These include two artificial neural network based techniques of cascade and Nonlinear-Autoregressive-Exogenous (NARX). The third considered approach is Gaussian Process Regression (GPR). The three forecasting methods are applied on electric load, temperature and humidity date for Sharjah city over 3 years, i.e, 2014–2016. The date of the first two years were used to train the model; while the third year was used to evaluate its forecasting capability. The results showed that all the models provide almost similar forecasting performance. However, ANN-NARX has a degree of advantage over other methods owing to its account for the time dependency.

In this paper a Genetic Algorithm (GA) augmented Convolutional Neural Network (CNN) procedure has been developed. The proposed approach was implemented in Python and applied on a simple three layer convolutional neural network to explore its effects on training the process. The global searching capability of the GA is exploited to initiate the training process of the conventional Back Propagation (BP) based CNN. The weights of the network are optimally initiated using the GA genetic algorithm rather than using random initializers before training. The proposed method of GA-BP based CNN shows better performance in terms of the training time and accuracy with respect to the conventional BP based CNN.

COVID-19 pandemic now affects the entire world and has a major effect on the global economy. A number of medical researchers are currently working in various fields to tackle this pandemic and its circumstances. This paper aims of developing a model that can estimate the number of deaths in the affected cases based on the documented number of older (above 65 years of age), diabetic and smoking cases. The Gaussian Process Regression (GPR) approach has been used to build the model and its performance was compared with a corresponding Artificial Neural Network (ANN) model. The model was applied to reliable data published by the World Health Organization (WHO) for different countries in North America, Europe and the Gulf region. The model provided impressive results with an excellent prediction of data from all the countries under investigation. The model may be useful in estimating the number of deaths due to any arbitrary number of inputs. It would also help to prepare effective measures to minimize the number of deaths.

In this paper, an efficient Support Vector Regression (SVR) and Artificial Neural Network (ANN) based IV models for Gallium Nitride High Electron Mobility Transistors (GaN HEMT) is proposed. The modeling approaches are applied on pulsed IV measurements of a standard size 1-mm GaN HEMT at four different ambient temperatures of 25, 40, 55, and 70°C. A non-parametric based Bayesian optimization is exploited to optimize the hyper parameters of SVR & subsequently improve its prediction capability. The two models were evaluated it terms of the mean squared error and their ability to accurately predict the IV characteristics. An excellent agreement with the measurement has been obtained over the entire operational regions of the transistor. Analytical formulas for the drain current based on both models have been provided and it can be easily implemented in standard CAD software.

This work addresses the increasing number of diabetic patients around the world. Neural network models have been developed and used to predict and classify the likelihood of a person to become diabetic. It presents three models based on neural network for the classification and prediction of diabetes. These models include a feedforward network, a pattern network, and a cascade forward architecture. The performance of the three models are compared in terms of accuracy, sensitivity, and specificity. All models are implemented and tested in MATLAB. The highest accuracy reported among the models is 91.1 %.

This paper presents an integrated artificial neural networks (ANNs) model with different global optimization techniques. The optimization techniques studied in this paper are the genetic algorithm (GA), the grey wolf optimization (GWO), the whale optimization (WOA), the dragonfly optimization (DA) and grasshopper optimization (GOA). These five techniques are compared based on their performance and time. All the techniques are tested on the same initial population. Moreover, the same experiment design (number of trials, iterations and individuals) is used in all the techniques. Among these techniques, ANN-GA and ANN-GWO show comparable and better performance than ANN-WOA, ANN-DA and ANN-GOA. The performance of ANN-GA can be improved by increasing the number of solutions and iterations. ANN-GWO showed a competitive performance among all algorithms with moderate number of solutions and iterations. ANN-GWO converges faster to solutions, which makes it more practical for faster and automatic processing. ANN-GA shows stable and robust performance with respect to ANN-GWO.

Impact of substrate thickness on self-heating effect for AlGaN/GaN HEMT is analyzed. Self-heating effect is analyzed using Technology Computer Aided Design (TCAD) simulation. GaN-HEMT with gate width of 1mm and gate length of 0.7 µm is considered for the analysis. In the simulation, substrate thickness is varied from 100 µm to 200 µm to study its impact on self-heating effect and drain current. In addition, the ambient temperature is varied from 300K to 700K and its impact is analyzed. Impact of gate-gate spacing on thermal performance is also analyzed. Furthermore, Kink effect on drain current at V GS =-2V is observed.

Diabetes affect many people around the world. While current treatments such as insulin injections and insulin pump devices are effective at managing diabetes, it still requires a manual effort that if not acted upon fast enough will lead to fatal results. This work designs and implements an artificial pancreas to improve the quality of life for diabetics by regulating the blood glucose levels and ease the burden of therapy for the insulin dependent. It presents a portable artificial pancreas system based on continuous glucose monitor (CGM), an insulin pump, and a microprocessor for implementing a control algorithm for automatic injection of the proper insulin dosage. The system also provides an efficient tool for monitoring the patient situation by submitting a regular report to the health care center and sending an alarm in case of any critical issue. The presented system is cost-effective compared to commercially available systems.

This paper introduces new analytical expressions that relate between voltage and current of a nonlinear two-port network. These formulas are defined and derived, analytically, from ${X}$ -parameters. They are also extension of the well-known linear two-port network parameters. These formulations are defined as follows. Linearized impedance parameters ( ${Z}$ -parameters) relate harmonic voltage components to harmonic current components at both network ports. Linearized admittance parameters are the inverse of linearized ${Z}$ -parameters and relate harmonic current components to harmonic voltage components. Linearized cascade parameters and linearized scattering transfer parameters relate entities at the input to that ones at the output of the network. Linearized hybrid parameters and linearized inverse hybrid parameters are defined in terms of a mixture of port variables. The proposed approach has been evaluated, and very good agreement has been obtained.

This paper presents an efficient parameter extraction method applied to GaN high electron mobility transistors. The procedure only relies on ${S}$ -parameter measurements at cold bias conditions to extract the extrinsic parameters of a 19-element small-signal model. Hybrid technique of particle-swarm-optimization and direct fitting has been developed and implemented. The extraction procedure has been optimized to consider measurements uncertainty and improve the reliability of the extraction. The procedure has been validated by multibias extraction for different device sizes. A very good agreement between simulations and measurements has been obtained.

Millimeter-wave technology is promising for the next generation of wireless communication. An appropriate channel model for typical scenarios in a built-up environment is crucially needed for development of future technologies such as 5G. This paper presents an efficient millimeter wave wideband propagation model that can be used for analysis and design purposes. Unlike the commonly used two-ray and statistical models, the proposed model has higher reliability and accuracy to simulate the considered environment. The uniform theory of diffraction has been used to calculate the total received signal due to multi-reflection–diffraction for both linear and circular polarization. Mathematical formulas have been derived for each considered ray contribution in terms of the building dimensions and the distances between the building, mobile and transmitter. The model has been used to evaluate the performance of the communication system at 28 and 73 GHz based on signal fading characteristics and power delay profile, and the simulation results have been compared with other published works

Sensing COVID-19, GOx (glucose oxidase enzyme) in exhaled breath condensate/saliva, bio-molecules like KIM (Kidney Injury Molecule) in human body and pH value in human body fluids have gained huge attention in the present scenario as well as in the past decade. Hence, for the first time, double channel technique in AlGaN/GaN High Electron Mobility Transistor (HEMT) is proposed and its applicability is demonstrated by biosensing application. Simulation using SILVACO Technology Computer Aided Design (TCAD) based on numerical solid state models has been extensively used for investigation and analysis. The sensitivity of double channel device is compared with single channel device and its performance is evaluated in terms of the transconductance. Unlike the single channel device, double channel device exhibited wide range of transconductance with respect to gate bias. The device recorded a sensitivity of 136%, which is 74% higher than the sensitivity of single channel device. Hence, it is inferred that the sensitivity enhances with the use of multiple channels and could be increased by increasing the number of channels. The results of this research show that the proposed sensor stands a promising candidate for future biosensing applications that demand high detection limits.

This paper presents conversion rules between ${X}$ -parameters, linearized impedance parameters, linearized admittance parameters, linearized cascade parameters, linearized scattering transfer parameters, linearized hybrid parameters, and linearized inverse hybrid parameters of a two-port network under large-signal operating conditions. The rules have been developed along with a set of equations that allow obtaining the expressions of each linearized parameter from the remaining ones. The proposed approach has been evaluated, and very good agreement has been obtained between calculated parameters and simulated ones.

Thermal effect represents a main challenge for GaN high electron mobility transistor (HEMT) especially for Sisubstrate based device under high power of operation. This paper addresses this issue by developing an efficient and simple feedback neural network (FNN) based electrothermal model. The model has been applied on a packaged GaN HEMT and implemented on advanced design software (ADS). The model showed a very good agreement with large-signal measurements at different classes of operation.

This paper reports the design and implementation of an Armstrong power oscillator for wireless power transfer (WPT) using GaN HEMT (high electron mobility transistor) at 4.6 GHz and GaN FET at 40 kHz. In the first design, a low power depletion-type device is used, simulated and integrated with a magnetically coupled resonant (MRC) WPT system using PCBs. The second design is based on a higher power enhancement-type transistor, where it is simulated and compared with the first design. High quality AC output waveforms with low distortion are obtained, which leads to low electromagnetic interference (EMI) levels that are required for sensitive applications such as biomedical technologies.

In this paper, four different equivalent circuit models to describe substrate/buffer loading effect in GaN HEMT on Si substrate have been investigated. The effect is characterized by Z-parameter measurements of open de-embedding structure for the considered device. A meta-heuristic optimization techniques of Grey-Wolf has been utilized to extract optimal values for the model elements. The models are validated by comparing their simulation results with the measured data. The extraction results are evaluated in terms of their accuracy and rate of convergence.

This work proposes a simpler automatic fault classifier that uses multi-layer perceptron (MLP) to identify faults in rotating machines. For classification, only statistical features and rotation frequency was taken in to consideration. Synthetic Minority Over-sampling Technique (SMOTE) was incorporated to handle imbalance classes in the machinery fault database. The proposed approach was evaluated on Machinery Fault Database (MAFAULDA) database. The proposed system has achieved an accuracy of 96.2%, which is comparable with the reported results in the literature. Results indicate that handling imbalance classes greatly increases the generalizability power of the MLP classifier on the machinery fault database.

Training artificial neural networks (ANNs) using global optimization techniques is becoming an attractive area of research. In this paper, combined modeling techniques of ANNs with genetic algorithm (ANN-GA), particles warm optimization (ANN-PSO) and grey wolf optimization (ANN-GWO) will be presented. The performance of these three techniques will be investigated in terms of efficiency and effectiveness for solving practical modeling problem. It has been found the recently developed ANN-GWO has a comparable performance with respect to the other techniques. ANN-PSO showed higher rate of convergence, which makes it more practical for real-time applications

This paper presents a framework for model order reduction based on Balanced Truncation (BT) method and Artificial Neural Networks (ANN). As social, architectural and technical models are witnessing rapid complexity increase, it has become essential to find solutions to obtain a lower complex version of models while still preserving the model integrity. The main idea of the proposed framework is to reduce uncertainty by cascading two different Model Order Reduction (MOR) approaches, the BT technique and the Jordan Recurrent Neural Network (JRNN). The reduced model obtained by the JRNN network is estimated using the Instrument Variable (IV) estimation approach. Finally, the BT reduced model and the estimated JRNN reduced model are cascaded to obtain the final reduced order version of the original plant. The cascading procedure takes the powerfulness of both techniques and present them in a single system capable to compensate for possible errors and uncertainties. Simulation results prove the efficiency of the proposed framework in terms of the deduction strength and the integrity level.

In this paper, a new accurate small- and large-signal equivalent-circuit-based modeling technique for gallium nitride (GaN) HEMT transistors grown on silicon substrate is presented. Despite X-parameters are developed as tools for the development of black-box modeling, they are used for equivalent-circuit-based model extraction. Unlike traditional modeling that uses the small-signal data to build with an indirect manner a nonlinear model, the proposed model is extracted from X-parameter measurements directly. However, similar to the equivalent-circuit-based models discussed in the literature, the new model is subdivided into extrinsic and intrinsic parts. The extrinsic part consists of linear elements and is related to the physical layout of the transistor. The intrinsic part can be extracted with the proposed analytical de-embedding technique. The nonlinear intrinsic elements are represented by new nonlinear lumped impedances and admittances whose extraction is carried out using a newly proposed technique. This new technique uses nonlinear network parameters, various X-parameter conversion rules, and basic analysis techniques of interconnected nonlinear networks. It is accurate and more advantageous than traditional transistor modeling techniques. The modeling procedure was applied to a 10 $\mu \text{m} \times $ 200 $\mu \text{m}$ GaN HEMT with a gate length of $0.25~\mu \text{m}$ . A very good accordance between model simulations and measurements was obtained, validating the modeling approach.

In this paper, an improved modeling approach has been developed and validated for GaN high electron mobility transistors (HEMTs). The proposed analytical model accurately simulates the drain current and its inherent trapping and thermal effects. Genetic-algorithm-based procedure is developed to automatically find the fitting parameters of the model. The developed modeling technique is implemented on a packaged GaN-on-Si HEMT and validated by DC and small-/large-signal RF measurements. The model is also employed for designing and realizing a switch-mode inverse class-F power amplifier. The amplifier simulations showed a very good agreement with RF large-signal measurements.

In this article, a simple, direct and reliable extraction method has been developed and applied to different sizes of Gallium Nitride (GaN) high electron mobility transistors (HEMTs). Instead of high-voltage gate-forward measurements with risk of device damage, the proposed approach uses only uncritical cold S-parameters at pinch-off (VGS

In this paper, a reliable small-signal model parameter extraction method for GaN high electron mobility transistor (HEMT) on Si substrate has been developed and validated with respect to different gate width devices. The main advantage of this approach is its accuracy and dependency on only pinched-off and unbiased ${S}$ -parameter measurements. The developed procedure shows reliable and physically relevant results for the investigated devices and scaled with the gate width. A very good agreement is obtained between small-and large-signal simulations and measurements of the considered GaN HEMTs.

In this paper, a reliable propagation model has been developed and applied to 5G systems to investigate their performance in a built-up environment. The model considers multiple reflections/diffractions of multi-path signals and provides an accurate prediction for the level of the received signal. Combination of ray tracing and uniform theory of diffraction (UTD) has been followed to derive a mathematical formula for the total electric field strength in terms of the model parameters (geometry, material, frequency and polarization). The model showed the expected results and a good agreement with other published papers has been obtained.

This paper presents a design of a 3.5 GHz GaN Low Noise Amplifier (LNA) using Microstrip interconnect technology on a gold plated PCB and using packaged commercial GaN-on-SiC HEMTs. Cascode configuration with inter-stage matching and independent biasing networks was used. The whole design was fully implemented using Microstrip except for the capacitors and was optimized to reduce the noise figure (NF) and improve the gain and stability. The LNA provided a gain of 13.5 dB with a noise figure (NF) less than 3dB from 2.9 to 3.7 GHz with power consumption 2.92W.

A novel reliable small signal model parameter extraction of asymmetric GaN-based Heterojunction Field Effect Transistors (HFETs) is proposed in this paper. An efficient systematic searching procedure based on pinch-off and cold weak-forward S-parameters has been developed to consider the asymmetric structure of GaN HFETs and find the optimal values of the gate and drain pad capacitances Cpg and Cpd. Considering that the depletion region extension is varied with the gate bias voltage V gs slightly even below pinch-off voltage, the obtained initial values of the extracted parameters are optimized by artificial bee colony (ABC) algorithm to improve the reliability of parameter extraction. The developed parameter extraction method shows excellent accuracy and reliability. The proposed procedure can be extended to asymmetric GaN devices with various process technologies. The developed approach has been validated by an asymmetric 0.15 μm GaN HFET over a wide range of bias conditions and frequencies., 2018 13th European Microwave Integrated Circuits Conference (EuMIC), September 23-25, 2018, Madrid, Spain

In this paper a small-signal and noise modeling approach is developed and applied to GaN HEMTs. The main advantage of the proposed method is its simplicity and dependency on only pinch-off S-parameter measurements. The achieved model shows good fitting with microwave experiments performed on nonlinear solid-state devices with different sizes. The model reliability is further confirmed by the observed scaling of the extracted parameters.