International Journal of Numerical Modelling-Electronic Networks Devices and Fields最新文献

This paper mainly designs the specific optimization process of antenna parameters or topology structure based on the online updated multi-layer perceptron (MLP) agent model, and introduces principal component analysis (PCA) technology to accelerate the overall optimization process. In addition, to address the problem of mixed optimization of machine learning assisted antenna parameters and topology structure, this paper proposes a new encoding approach, which enables two forms of variables to jointly participate in antenna optimization. The specific approach is to treat the topology structure and antenna parameters as a state and encode them in binary, thereby unifying them into a set of binary variables, which facilitates the introduction of subsequent machine learning and optimization algorithms. By optimizing an ultra-wideband G-patch monopole antenna and a microstrip pixel antenna, it has been demonstrated that the proposed method has better optimization performance compared to traditional optimization methods and can be applied to antenna parameters or topology structures.

This study introduces a new formulation of the Wave Concept Iterative Process (WCIP) method in elliptic coordinates to analyze electromagnetic scattering from an infinitely long, inhomogeneous, multilayered elliptical cylinder. The foundation of this approach lies in the definition of emitted and reflected waves at the interfaces of dissimilar media, utilizing the tangential components of the electric field and current density. The unknown waves resolved from a set of two equations. The first equation, formulated in the spectral domain, characterizes the response of the environment around the structure. The second equation guarantees the continuity of the electromagnetic fields and enforces boundary conditions at the interfaces that separate different regions. The system of equations is resolved by an iterative process where a discrete angular Mathieu Transformation Algorithm is introduced to toggling between the spatial and the modal domain. The proposed approach is suitably implemented in a specific computational code to achieve a numerical solution. Some numerical results are presented, involving a structure composed of metallic strips positioned at the interfaces of four elliptical dielectric layers, which is illuminated by a transverse magnetic (TM) plane wave at an oblique angle of incidence. The result is in agreement with those published in the literature.

This article presents a comprehensive temperature-dependent analysis of Ferroelectric Junction-less Surrounding-Gate FETs (Fe-JL-SG FETs). To evaluate the effectiveness of ferroelectric integration, a detailed comparison between Fe-JL-SG FETs and conventional Junction-less Surrounding-Gate FETs (JL-SG FETs) is carried out. Key device parameters—including surface potential, electric field, electron concentration, electron velocity, transconductance (g_m), output conductance (g_d), capacitance (C_GG), cutoff frequency (f_t), and I_ON/I_OFF ratio—are systematically investigated over a broad temperature range (100–500 K). The results demonstrate that Fe-JL-SG FETs exhibit superior electrical and analog performance with significantly reduced sensitivity to temperature variations compared to JL-SG FETs. This enhanced behavior is attributed to the negative capacitance effect introduced by the ferroelectric material, which effectively reduces the subthreshold-swing (SS) below the Boltzmann limit (60 mV/decade). However, this effect weakens as temperature increases. The simulation framework, developed using the ATLAS 3-D device simulator, exhibits strong congruence with the proposed analytical model, thereby validating the theoretical constructs and reinforcing the predictive reliability of the developed approach.

This research presents the design and optimization of a miniaturized frequency-selective surface (FSS)-embedded fractal MIMO antenna for 28 GHz 5G New Radio (NR) applications. The compact antenna, measuring 4.2 × 1.87 λ, employs a 7 × 3 unit-cell array to achieve superior radiation characteristics. The design methodology integrates fractal geometry analysis, FSS loading for electromagnetic enhancement, and machine learning-based optimization to achieve optimal gain, impedance bandwidth, and radiation efficiency. The antenna, realized on a Rogers RT5880 substrate, is optimized using Random Forest (RF), Artificial Neural Network (ANN), and regression-based predictive models. Performance evaluation using R², MAE, and MSE confirms the robustness of the optimization framework. Key design challenges include maintaining high isolation among compact MIMO elements, ensuring stable broadband response, and balancing miniaturization with gain without sacrificing efficiency. These challenges are addressed through precise geometric refinement, effective FSS coupling, and iterative machine learning tuning. Experimental validation using a ZVA 67 vector network analyzer demonstrates a broad impedance bandwidth of 2200 MHz (26–28.2 GHz), a peak gain of 7.53 dBi, and a radiation efficiency of 88.49%. The FSS integration enhances isolation, suppresses surface waves, and improves overall mm Wave performance. The proposed ML-assisted FSS-MIMO design provides an efficient solution for next-generation 5G wireless communication systems.

The rapidly developing field of 3D integrated circuit technology offers a revolutionary solution to the increasing number of transistors on a chip: stacking numerous silicon layers vertically. Stacking layers greatly lowers latency and energy usage in 3D integration. Unlike the conventional 2D NoC, the three-dimensional Network-on-Chip (NoC) paradigm offers a multitude of options and difficulties, piquing researchers' attention. For 3D NoC systems, thermal control is a more important consideration than for 2D ones. This research presents an effective XYZ routing method for thermal control. The suggested approach uses a signal produced by a basic controller to dynamically modify traffic flow. By adjusting the XYZ routing technique, this signal prolongs the life of the device by preventing thermal hotspots in the 3D NoC layers. Experimental results show that the suggested method can reduce layer temperatures while preserving performance levels comparable to the original XYZ routing algorithm in comparison to earlier efforts. Utilizing VHDL to construct the design, the 3D NoC design also demonstrates an improvement in space efficiency by requiring fewer LUTs on the FPGA than previous designs.

In this paper, analytical models have been developed for the analysis of flicker and thermal noise in a double-gate junctionless MOSFET with double metal gates (DMDG). We first formulate drain current and bulk charge density models of our proposed device structure. The drain current as well as the noise models are validated with experimental results, and excellent matching has been observed. The effect of temperature on the noise power density spectrum is also studied. The modeling results show that flicker noise reduces with an increase in the length and thickness of the channel. This is due to the reduced influence of traps and fluctuations on carrier mobility. The flicker noise changes inversely with V_gs, as it increases with an increase in V_ds and temperature. The thermal noise decreases when the thickness and doping concentration of the channel, along with the gate-oxide thickness, are increased. This results from more ionized impurity scattering and weaker gate control over the channel. Also, the thermal noise enhances with the increase in V_gs, whereas it varies inversely with V_ds. The proposed model can be helpful to analyze the influence of noise on the DMDG device characteristics, which has potential applications in analog/RF circuits.

This paper presents a data-driven methodology for designing a highly efficient power amplifier (PA) incorporating spoof surface plasmon polariton (SSPP) based matching networks. The PA employs a CG2H40010F GaN HEMT transistor and is optimized to operate across 3.0–4.8 GHz. Log-periodic line (LPL)-based SSPP unit cells are utilized in both the input and output matching networks to achieve enhanced field confinement and circuit miniaturization. A two-stage design approach is adopted: initial optimization is performed through harmonic balance simulations using simplified real frequency technique (SRFT), followed by electromagnetic (EM) co-simulation to capture full-wave effects. A Bayesian Optimization framework is applied to systematically tune the SSPP geometrical parameters and LPL configurations, using output power (P_out), power-added efficiency (PAE), and gain as optimization objectives. The final design demonstrates a peak P_out of 43 dBm, a drain efficiency exceeding 80%, and a gain of 13.6 dB. The proposed approach enables an efficient and compact SSPP-based PA design without requiring hardware measurements, offering a promising direction for future wideband front-end amplifier solutions.

This research addresses instability in GaN HEMT half-bridge circuits caused by high switching speeds and parasitic parameters, which induce oscillations at the gate-source and drain-source terminals. These oscillations increase switching loss, risk false triggering, and can lead to device failure. A small-signal model based on a bootstrap driver is proposed to study this phenomenon. During dead time, reverse conduction and parasitic-induced positive feedback worsen the oscillation. The system is modeled as a second-order underdamped network, and a damping ratio is introduced to evaluate stability. Strategies such as reducing common source inductance and optimizing gate resistance are used to improve damping. Additionally, an RCD circuit is added across the gate-source to absorb resonant energy and suppress negative voltage spikes during turn-off. LTspice simulations and pulse testing using GS66508B devices show a reduction in negative spikes from −3.35 to −1.20 V and in turn-off loss from 8.23 to 7.68 μJ. These results verify the effectiveness of the proposed method in enhancing switching stability.

The recent developments in wireless technology have given rise to the emergence of wearable antennas (Ants.). These Ants. are employed in the context of wireless body area networks (WBANs), which find applications in diverse fields including healthcare, military operations, sports, and identification systems. In this work, an ultra-wideband, compact, low-profile, and low specific absorption rate (SAR), metamaterial (MTM) integrated wearable Ant. featuring a flexible for WBAN applications is presented. The Ant. and MTM structures are designed on a felt substrate with sizes of 50 × 50 × 1 mm³ and 61.60 × 61.60 × 2 mm³, respectively. The proposed MTM Ant. has a physical thickness of only 4.14 mm, offering the thinnest profile among similar MTM-integrated UWB wearable Ants. reported in the literature. The MTM structure is designed to mitigate the SAR effect and enhance the Ant. performance parameters, including impedance matching, radiation pattern, and realized gain. The performance of the proposed MTM Ant. was evaluated through simulations conducted in free space and on a human body model, specifically on the chest, arm, and leg. The SAR values of the MTM Ant. are found to be well below the maximum permissible limits of 0.49 W/kg (1 g) and 0.125 W/kg (10 g) established by the European and US standards. Additionally, the integration of an MTM structure into the Ant. configuration was demonstrated to result in a notable enhancement in the simulated peak gain of the Ant., reaching 6.95 dBi. Furthermore, it was observed that the maximum front-to-back ratio exhibited an increase to 24.78 dB. Also, the bending conditions of the MTM integrated Ant. are evaluated in detail. The designed Ant. and MTM were fabricated and subsequently subjected to measurements. The results of the measurements were found to be in good agreement with the results of the simulations. Simulation and measurement results show that the proposed Ant. with MTM has satisfactory performance characteristics for WBAN applications.

This study presents a hybrid numerical technique for solving multi-dimensional time-fractional wave equations, combining Laplace transforms and a Chebyshev-node-based Lagrange pseudo-spectral method (LPPSM). The method is designed to overcome the computational burdens of time-stepping schemes for fractional derivatives. By transforming the temporal derivatives into the frequency domain, it enhances numerical stability and avoids time-stepping constraints. The spatial discretization via LPPSM ensures spectral convergence. The time-domain solution is recovered using an optimized Talbot contour inversion. The efficacy of the proposed method is demonstrated on benchmark problems, including the fractional Klein-Gordon equation, achieving exceptional accuracy (errors on the order of $��{10}^{�-�12�}�$). Crucially, to bridge the gap with practical applications, the method is applied to a time-fractional telegraph equation modeling signal propagation in a transmission line with physical parameters $��\left(�R�,�L�,�C�\right)�$. The results confirm the method's high computational efficiency, superior accuracy, and potential for real-time analysis of fractional-order dynamics in high-frequency electronic systems.