This study introduces a compact four-port UWB MIMO antenna featuring dual-band rejection capabilities aimed at suppressing interference from coexisting wireless services, specifically WiMAX at 3.5 GHz and WLAN at 5.5 GHz. The antenna employs an inverted C-slot etched into the radiator to achieve the WiMAX notch, while EBG structures are integrated to enable suppression of the WLAN band at 5.5 GHz. Fabricated on a low-cost FR4 substrate with dimensions 52 × 52 × 1.5 mm³ (ε_r = 4.5), the proposed design achieves high port isolation exceeding 15 dB across the 3.1-10.6 GHz UWB range. Simulated results show an operational bandwidth from 3 to 11 GHz, extending beyond 12 GHz in measurements, without the need for additional filters or decoupling structures. The antenna exhibits quasi-omnidirectional radiation patterns with a peak gain of 7.2 dBi and significant gain suppression at the notch frequencies (-1.5 dBi at 3.5 GHz and -1.2 dBi at 5.5 GHz). It maintains a VSWR below 2 across the UWB and achieves radiation efficiency above 90% outside the notched bands. The envelope correlation coefficient remains below 0.005, enabling a high diversity gain approaching 10 dB. The EBG structures effectively reduce mutual coupling, allowing a compact element spacing of just 2 mm (approximately λ/12.5 at 12 GHz). Both simulation and measurement results validate the proposed design's suitability for mitigating co-channel interference in UWB-MIMO applications, including satellite communications in the S/C/X bands and high-speed wireless systems.

This paper presents the design and implementation of a compact 4-port antenna element with high isolation for 5G sub-6 GHz applications. Four V-shaped patch elements are arranged orthogonally on a 1.58 mm thick FR4 substrate to mitigate mutual coupling in the proposed structure. A defected ground plane method is utilized to further enhance and optimize the characteristics of the antenna at the operating frequency. The antenna operates in the 3.15-4.1 GHz frequency range, providing a 950 MHz impedance bandwidth at -10 dB, making it suitable for mobile terminals within the 5G sub-6 GHz band. The orthogonal polarization results in isolation levels below -18.1 dB, making the antenna ideal for 5G handset communications. This high isolation is reflected in an envelope correlation coefficient (ECC) of less than 0.04, while the diversity performance is verified by a total active reflection coefficient (TARC) of less than -10 dB. The channel capacity loss (CCL) of the four-port antenna element is calculated to be below 0.1 bps/Hz at 3.5 GHz. The MIMO antenna was fabricated, and its measured performance closely matches the simulated results, confirming that the proposed MIMO antenna is well-suited for future sub-6 GHz cellular communications.

We present an eight-page in-depth study of a single-layer broadband reflectarray antenna operating over the 8 GHz to 12 GHz X-band. The array employs a dual-ring hex-slit unit cell and a physics-informed deep-learning (DL) surrogate model that reduces geometry optimisation time by ×120 compared with brute-force sweeps. The 30 cm × 30 cm prototype comprises 273 passive elements, delivers a 530° reflection-phase span, 27 dB peak gain, 56% aperture efficiency and 34.6 dB cross-polar discrimination. A residual network trained on 5000 HFSS datapoints predicts reflection phase with 0.9° MAE, whereas its inverse sibling outputs element radii in 10 ms. CST full-wave simulations and a preliminary S-parameter measurement corroborate the synthesis accuracy to within 0.25 dB. Comprehensive parametric, angular-stability and computational analyses provide guidance for extending DL-assisted reflectarrays to higher frequencies and reconfigurable architectures.

Severe pneumonia poses a significant threat to public health. Delayed diagnosis is a core challenge in treatment. This study uses two rapid, low-cost spectroscopic fingerprinting techniques - Raman spectroscopy and attenuated total reflectance Fourier transform infrared (ATR-FTIR) absorption spectroscopy - to analyze biofluids such as blood and bronchoalveolar lavage fluid (BALF). In contrast to our earlier work which combined infrared spectra with clinical biochemical test results, this paper focuses solely on the spectral data to validate a fast and label-free diagnostic method. We used a spectral transformer network (STNetwork) to perform AI-based classification of severe pneumonia from the spectral fingerprints of blood and BALF. While both modalities are effective, FTIR spectroscopy exhibits superior diagnostic precision (97.78% test accuracy) and stability (SD < 0.0139) for blood samples. BALF offers a unique window into the local lung microenvironment, and both metabolomic analysis and spectral fingerprint classification were performed. The classification results for BALF Raman spectra (enhanced with surface-enhanced Raman spectroscopy) gave a training accuracy of 96.71%±1.86% and a testing accuracy of 90.62%±3.95%, better than the classification results for BALF FTIR spectra. The present study provides a reliable technical foundation for developing rapid and high-accuracy screening solutions for severe pneumonia.

This paper introduces a deep learning assisted sensor for material classification based on two adjacent split-ring resonators unit cell sensor. This sensor operates in the frequency range from 7 GHz to 8 GHz. The sensor is designed to differentiate between different dielectric materials based on their reflection and transmission properties. A dielectric container is used to hold different samples. Reflection and transmission coefficients for different materials are used for classification between different dielectric materials. These materials are also characterized by using Dielectric Assessment Kite (DAK) for verification with the proposed method. The Dual Split Ring Resonator (DSRR) unit cell enhanced resonance characteristics facilitate the classification process for distinguishing different dielectric materials. The measured results of the proposed sensor exhibit a broad detection range, accurately identifying various samples based on their unique resonant frequency responses. The proposed sensor finds utility in industrial applications, identifying and categorizing various different dielectric materials. In addition, the proposed design is used to measure a mixture of two different materials with different volume mixing ratios. The measured samples are used to train a convolution neural network to predict the mixing ratio from the measured S-parameters. The combination of this sensor and the trained model is found to be an efficient tool that determines the mixing ratios of different samples in a fast way. This concept can also be useful to be applied on other types of sensors and other sensing parameters.

This paper proposes a novel and compact conformal fork-shaped microstrip patch antenna operating at 10.2 GHz for X-band wireless communications. It features a fork shaped unique radiating structure optimized for impedance matching, bandwidth enhancement, and compactness. Parametric studies confirmed optimal performance at feed width 3.5 mm and thickness 0.4 mm. An operating band from 9.31 to 11.88 GHz for S₁₁ <= -10 dB with a bandwidth of 2.57 GHz is attained. It exhibits a peak gain of 5.6 dB at 10.2 GHz with radiation efficiency 88.29%. To validate its suitability for flexible and modern wireless applications, conformal models are developed, and their performance is analyzed for metrics like operating band, radiation patterns, peak gain, and radiation efficiency. It is prototyped on a Rogers RT Duroid 5880 substrate, and experimental validation demonstrates strong correlation between simulated and measured characteristics.

In this paper, a high-efficiency, center-fed dynamic seawater monopole antenna is proposed. The antenna's radiation efficiency increases by raising the feed point position, adding a metal tube above the feed point, and reducing the inner diameter of the water pipe below the feed point. Concurrently, the effects of the feeding position, the length of the metal pipe and the inner diameter of the water supply pipe on the current distribution are analyzed by theoretical modeling. FEKO electromagnetic simulation software is applied to simulate and analyze metal tubes with different lengths and water supply side pipes with different inner diameters. The simulation results indicate that the antenna attains optimal performance when being configured with a water supply pipe of 1 cm inner diameter and a 30 cm metal tube, achieving a maximum gain of 0.93 dBi and a peak radiation efficiency of 52%. Based on the simulation data, a simplified center-fed dynamic seawater antenna prototype is designed and fabricated. Experimental validation confirms that the seawater serves as the primary radiating element. The measured radiation characteristic curves exhibit consistent trends with the simulated results.

This article proposes a miniaturized dual notched ultrawide bandpass filter (BPF) for ultra-wideband (UWB) indoor applications. The initial operational spectrum recognition is realized through the resonances of multiple mode resonator (MMR). Then both the passband and stopband characteristics are improved substantially by mounting distinctly shaped meander resonators cascaded with open loop ring resonator on the MMR. Further, the interdigital coupled lines are also meandered to contribute in filter size reduction along with tightening the coupling between the effective filter structure and input/output ports. The elimination of interfering signals within the passband caused by C-band satellite downlink and fixed satellite service uplink is facilitated by two sharp notches at 3.76 GHz and 6.82 GHz frequencies. Concurrently, this miniaturized filter is also characterized by its wide passband of 6.42 GHz with fractional bandwidth (FBW) 110.88%, good selectivity of 0.85, minimal insertion loss differing between 0.44 dB and 0.85 dB, wide upper stopband of 5.11 GHz, etc. ensuring its suitability as a practical UWB filter. The design is fabricated and measured to compare with the simulated outcomes and validated by the obtained resemblance between the measured and simulated filter outputs.

A broadband efficient reflective linear polarization converter has been developed using a metasurface. The converter consists of a U-shaped array integrated with cross dipoles. The U-shaped unit facilitates cross-polarization conversion, while the cross dipole serves to broaden the operational bandwidth. It has been demonstrated that the proposed converter can transform an incident linearly polarized wave into its orthogonally polarized reflected wave across a broadband frequency range of 18.94-51.03 GHz with a conversion efficiency exceeding 90%. The efficiency remains above 90% even as the incident angle increases to 45°, within the frequency ranges of 19.18-20.19 GHz, 26.01-30.93 GHz, and 48.8-49.77 GHz. The strong electric and magnetic responses at multi-resonances reveal the broadband polarization conversion mechanism. A prototype was fabricated to verify the performance. Measured results exhibit good consistency with the simulated ones.

In this paper, a novel, compact, broadband circularly polarized (CP) semi-hexagonal monopole antenna is proposed. The antenna integrates a rectangular open loop and nonuniform asymmetric ground planes. The proposed design is fabricated on a 1.6 mm thick FR4 substrate with overall dimensions of 0.83λ_c × 0.70λ_c × 0.021λ_c at the center frequency f_c = 3.935 GHz. It achieves an impedance bandwidth of 4.51 GHz (1.68-6.19 GHz, 114.63%) and an axial ratio bandwidth of 2.6 GHz (2.3-4.9 GHz, 72.3%). The antenna offers realized gains of 1.4 dBi and 0.5 dBi, with corresponding efficiencies of 64% and 51% at 2.4 GHz and 3.5 GHz, respectively. It exhibits right-handed circular polarization (RHCP) across the ISM and WiMAX frequency bands. Furthermore, the antenna's performance is validated in a realistic environment using an Arduino-based wireless voltage monitoring system, demonstrating reliable data transmission with minimal loss. These results confirm the antenna's suitability for IoT applications requiring stable and efficient wireless communication.

A novel single-cavity triple-mode combined-type rectangular patch resonator (CRPR) is proposed in this paper, which is realized by integrating the rectangular patch structure with the rectangular substrate integrated waveguide (RSIW) structure. By cleverly designing the length-to-width ratios of both the RSIW structure and the patch structure, as well as the size ratio between them, the three higher-order modes of the CRPR can be resonance. Then, a highly selective bandpass filter (BPF) is realized through a special feeding structure. To demonstrate the method, an instance of a BPF is designed, synthesized, fabricated, and measured. The consistency of all results validates the effectiveness of the proposed design method. The proposed filter offers advantages such as relatively compact size, easy integration, and high selectivity.

This paper describes how the low sidelobe pattern synthesis of planar arrays with a distorted triangular or rectangular lattice, caused by row displacements, can be improved using the iterative Fourier transform (IFT) method. Array antennas with a rectangular, or triangular lattice combined with row displacements have an array factor that lacks periodicity in cosine u-v space for the u-direction. This means that for the u-direction, the pattern synthesis using the IFT method is limited to far-field directions that belong to the rectangular sector of the array factor computed by the inverse 2D FFT. Missing far-field directions in the pattern synthesis occur when the width of the computed array factor (AF) in u-v space is <2. In this case, not all far-field directions in visible u-v space are engaged in the pattern synthesis. The solution to this problem is to reduce the inter-element spacing along the rows with a factor two by including dummy elements with zero excitation between the active elements in each row. In this way, the width of AF, computed by the inverse 2D FFT, is doubled in u-v space. This doubling will result in twice as many far-field directions in the u-direction being involved in the pattern synthesis. After successful synthesis, all dummy excitations are removed from the synthesized set of excitations. The element excitations thus obtained, without the dummy ones, still perform the same as the original excitation obtained from the pattern synthesis. Three examples will demonstrate the validity of this solution.

This study investigates the relation between the physical parameters and the scattering parameter (S₁₁) curves of antennas, and proposes two deep-neural-network-based frameworks respectively for antenna forward and inverse designs, improving the design efficiency compared to the conventional electromagnetic (EM) simulation approaches. In this study, a one-dimensional (1D) U-Net is utilized as the backbone of the two models and is enhanced with multiple mechanisms - the diffusion mechanism, channel attention, and spatial attention. Therefore, the models more effectively capture the sequential features of data. In the forward design, the model quickly predicts the S₁₁ curves from given physical parameters with an accuracy improvement of at least 63% RMSE and 70% MAE compared to the improved one-dimensional convolutional neural network (1D-MCNN) and deep multi-layer perceptron (DMLP), thus realizing the surrogate model of conventional methods to some extent. In the inverse design, another model directly infers the physical parameters corresponding to the target S₁₁ curves with an accuracy improvement of at least 21% RMSE and 38% MAE compared to the baseline models (1D U-Net and MLP), thereby eliminating the iterative process of traditional methods and accelerating the antenna design. The experimental results demonstrate the significant advantages of the proposed deep neural network frameworks in terms of accuracy and efficiency for both forward and inverse designs of antennas, offering a powerful alternative to conventional electromagnetic simulation-based approaches.

This paper presents a varactor-tuned low-pass filter (LPF) with high sharpness-factor and steep stopband suppression at different tunable frequencies using defected ground structure (DGS). By periodic loading four series-parallel capacitive compensation DGS units with high quality factor Q and the varactor diodes in shunt, a central filter is formed. To suppress the spurious passband and improve the rejection of stopband, four extra units in the form of split-ring stepped-impedance DGSs also with high Q factor were introduced and loaded at both ends of the central filter. The prototype of the proposed LPF is designed, fabricated and measured. Simulation and measurement results exhibit a good agreement. The measured results demonstrate a continuous tuning range of 12-13.5 GHz for the cut-off frequency, with an insertion loss in the passband better than 0.8 dB and a sharpness factor less than 1.08 across the entire range. The stopband rejection level is better than 30 dB and can be extended up to 40 GHz.

Early detection of breast tumour is crucial for reducing the likelihood of mastectomy. To monitor the dielectric changes in breast tissue caused by the formation of tumorous cells, a novel biocompatible implantable antenna sensor is proposed. This flexible implant, measuring just 5 mm × 5 mm × 0.25 mm, operates in the ISM band at 2.45 GHz for real-time breast tumour detection. It is wirelessly powered via Wireless Power Transfer (WPT) operating in the mid-band range of 1.2-1.4 GHz. The antenna achieves an ultra-compact volume of 12.5 mm³ through closed-loop structures and meandered strips that enhance radiation efficiency. Inside abnormal breast tissue with a relative permittivity (ε_r) of 52.7, the antenna demonstrates a reflection coefficient of -17 dB and offers a -10 dB bandwidth of 330 MHz. The sensor is activated when the tissue permittivity rises above 15, achieving a maximum gain of -10 dBi. The antenna has been fabricated, and the simulated results have been validated in-vivo. This design enables proactive detection of tumour cell formation within breast tissue, allowing treatment before it spreads. It is particularly suitable for individuals with a genetic predisposition to breast tumour, offering continuous monitoring for early intervention.

Reconfigurable Intelligent Surface (RIS), as one of the potential key technologies for 6G, can effectively solve the problem of millimeter-wave links being obstructed by constructing an intelligent and controllable wireless communication environment. In this paper, a joint beam tracking algorithm based on RIS selection is proposed for the scenario of multi-RIS-assisted millimeter-wave vehicle-to-infrastructure (V2I) communication. The aim is to select as few RISs as possible to aid communication while the performance of beam tracking can be maximized. Firstly, the beam tracking model jointly composed of line-of-sight paths and virtual line-of-sight paths constructed by multiple RISs is derived based on the multiple-input-multiple-output model in a 3D road scene, and the beam tracking under this combined path is realized based on the Extended Kalman Filter (EKF) algorithm. Second, for the RIS-assisted millimeter-wave V2I scenario, a new metric to quantify the beam tracking performance is comprehensively designed based on the received signal-to-noise ratio, the beam angle variation, and the distance variation from the RIS to the vehicle. Finally, based on this metric, the joint beam tracking is realized by the RIS selection strategy and the EKF algorithm under the combined path. Simulation results show that the joint beam tracking algorithm based on RIS selection proposed in this paper has lower beam tracking error than the traditional signal-to-noise ratio based beam tracking algorithm.

Traditional ways of measuring compaction of asphalt, which involve destructive coring, are labor-intensive, time-consuming, and cause permanent damage to the road. This paper presents a non-destructive alternative using a dual-ring resonator sensor (DRRS) integrated with a Vector Network Analyzer (VNA) to evaluate asphalt compaction. The sensor design takes advantage of the electric field that forms between the first and second rings. This field can penetrate the asphalt layer to a depth of up to 50 mm and responds to changes in compaction levels. By putting asphalt samples of different densities on the sensor and measuring scattering parameters (S-parameters), changes in the resonant frequency are shown. These shifts were correlated with asphalt's physical properties through empirical equations. The results showed that the resonant frequency and the reflection coefficient (S₁₁) were -25.5 dB and 1.38 GHz, respectively, at a 75% compaction level. The frequency changed to 1.17 GHz at 100% compaction, and S₁₁ was -17.6 dB. Increasing the compaction of asphalt makes the air gaps in the material smaller, which makes its permittivity higher. Calibration was performed to mitigate the influence of temperature on permittivity measurements, thereby improving compaction. Overall, this method provides a fast, precise, and non-destructive way to check the quality of asphalt, significantly enhancing road construction and maintenance processes.

Pseudo-direct-drive (PDD) machine is a new type of permanent magnet machine with high torque density and efficiency. PDD with consequent poles can reduce the influence of outer PM on electromagnetic torque, but it has the disadvantage of high eddy current loss which will limit the range of speed. By transferring PMs from high-speed rotor to low-speed rotor, the eddy current loss in PMs is reduced, and the high-speed rotor is more robust. In this paper, a flux-switching consequent pole PDD (CP-PDD) machine is built. After optimization through a multi-objective genetic algorithm, the superiority of the proposed machine to regular CP-PDD is demonstrated by comparing it through the finite element method. The output torque of the proposed machine is greatly affected by the direct drive torque. A prototype is built and tested to verify the proposed machine. Results show that the proposed machine is more suitable for high-speed operation due to the reduction of loss and robustness of the high-speed rotor. The working temperature of the proposed machine is analyzed, and there is almost no irreversible demagnetization.

Hybrid excitation starter generator (HESG) has an increased number of magnetic potential sources, leading to issues such as complex magnetic circuits, numerous structural parameters, and low space utilization. These factors cause traditional analysis methods to have long cycles and low accuracy. In this paper, a new type of salient pole electromagnetic and permanent magnet composite pole HESG is proposed, and an analysis method combining hierarchical optimization and Taguchi method to analyze the influence of different structural parameters of composite pole rotor on the HESG performance is proposed. Response surface method was used to simulate the stator groove with multiple objectives, analyze the electromagnetic characteristics of HESG, complete the performance optimization, prototype, and conduct experiments. The results show that the amplitude of the air gap magnetic density base of HESG is increased by 7.4%; the distortion rate is reduced by 10.6%; the output voltage is increased to 127.68 V; the output performance and magnetization ability are significantly improved; and the overall performance of the HESG is improved.

Magnetic forces play a significant role in modern engineering applications, from medical imaging, data storage to transportation and industrial machinery. Accurate and efficient computational methods for magnetic force are necessary for engineering design and optimisation. However, different methods are typically based on distinct assumptions and are suited to different application scenarios. To assist researchers and engineers in selecting the most appropriate method for their specific needs, this review provides a comprehensive overview of various numerical approaches for calculating magnetic forces across different magnetic systems. Several key methods such as The Dipole Method, Filament Method, Finite Element Method (FEM), Energy Method, Maxwell Tensor Method, Integral Method and Boundary Element Method (BEM) are discussed in detail, demonstrating their fundamental theories, applicable scenarios, advantages, and limitations. Recent advancements and improved versions of these methods are also covered, demonstrating their enhanced accuracy and efficiency. In addition, the potential solutions of these methods and future directions of developing advanced magnetic force computation techniques are also discussed in this paper.