This study introduces a machine learning (ML)-based method for point-of-care (POC) colorimetric testing of total antioxidant concentration (TAC) in saliva, an important biomarker for health monitoring. The approach leverages ML to accurately classify color intensity in the POC test. Saliva samples were collected and imaged at specific intervals during the colorimetric reaction, generating a dataset representative of various antioxidant levels. Four classifiers (Convolutional Neural Network, Support Vector Machine, K-Nearest Neighbors, and Single-layer Feed-Forward Neural Network) were evaluated on distinct datasets, with Convolutional Neural Network (CNN) consistently achieving superior performance. To enhance classification accuracy, stacking-based ensemble learning was applied, combining CNN predictions with a Support Vector Machine (SVM) meta-classifier, achieving up to 92% accuracy. Additionally, YOLOv4-tiny was utilized for object detection to isolate regions of interest in the images, creating a refined dataset that a CNN model is then classified with ca. 98% accuracy. This YOLOv4-CNN approach not only improved accuracy but also simplified the model architecture. The integrated object detection and CNN models were deployed on an Android application, enabling real-time TAC analysis on a smartphone with 98% accuracy and a fast readout time of 2 minutes. This method offers a robust, efficient, and accessible solution for POC antioxidant testing.

In this paper, a novel H-slot resonator of size 30 mm x 20 mm x 1.56 mm for fracture detection, backed with a perfect electric conductor (PEC), designed at 2.54 GHz is presented, and its performance is evaluated. Concurrently, an equivalent circuit model of the resonator is developed, and its performance coherently agrees with the CST model. The design is tested using a hybrid tissue phantom based on a second-order Debye dielectric tissue model. The detection of fractures of several thicknesses with a minimum width of 2 mm and a maximum of 10 mm was compared. Overall, the proposed design improves the detection of fractured regions in a bone with a 2 mm crack width as the smallest detectable crack size.

Polaritons, part-light−part-matter waves, enable the control of light at the subwavelength scale. Interfacial behaviors play a critical role in polariton manipulation, with negative refraction showing promise for high-resolution focusing. However, reflections pose a substantial challenge, especially in applications where backscattering is unwanted. To address this issue, we propose a structure composed of planar junctions of metasurfaces, each supporting in-plane hyperbolic polaritons with misaligned optical axes. We demonstrate that when the asymptote of the incident hyperbolic isofrequency contours (IFCs) aligns with the interface normal, the reflected waves transform into highly lossy one-dimensional ghost polaritons (HL-1DGPs), channeling energy near the interface. The refracted waves also convert into HL-1DGPs when the outgoing IFC asymptote aligns with the interface normal. Leveraging these phenomena, we design polaritonic lenses and absorbers with greatly reduced reflection. These insights into the interfacial behaviors of hyperbolic polaritons under symmetry breaking have implications for creating polaritonic elements beyond the diffraction limit.

To increase the detection dynamic range of slope-assisted Brillouin optical time-domain analysis (SA-BOTDA) system, we propose a configuration using unbalanced frequency-shifted stokes and anti-stokes sidebands as continuous probe light simultaneously to expand the region of effective stimulated Brillouin scattering (SBS) spectrum existed in frequency domain. The proposed scheme fully utilizes unbalanced double sidebands' gain, loss and corresponding phase spectra, constructing linear regions by specific data processing methods and has pump power-independence characteristic. Besides, scheme of dual-frequency agile change is employed to broaden the linear region at the cost of detection speed. The dynamic detection range of the proposed system can be increased to over 180 MHz, with a spatial resolution of 3.5 m and 500 MHz sampling rate for vibration detection.

The second near-infrared region (NIR-II, 900-1880 nm) spectral window has garnered significant attention in bioimaging due to its moderate light absorption, diminished photon scattering and reduced autofluorescence. Exploiting NIR-II fluorescence, confocal microscopy has achieved deep in vivo imaging. In this study, we have identified that the fluorescence with wavelength beyond 1400 nm offers superior imaging quality for NIR-II confocal microscopy, irrespective of the laser excitation source being continuous-wave or pulsed. Furthermore, leveraging the multiphoton excitation capabilities of femtosecond laser, we have successfully integrated multiphoton excited visible fluorescence channels into the NIR-II fluorescence confocal microscopic system. We have successfully employed this novel system to acquire up to six distinct fluorescence microscopic imaging channels with negligible cross-channel interference, as well as multi-channel and large-depth in vivo observation of mouse brain and kidney.

In this article, body scanning and imaging using a triangular microstrip antenna (MSA) with microstrip line feeding is presented. The resonant frequency of this antenna is 2.383 GHz having 20 dB bandwidth 3 MHz. The results comply with the 2.36 to 2.39 GHz band that the Federal Communication Commission (FCC) has designated for medical related applications. This antenna is used in scanning the human body model to detect the presence of tumor. The scan results are used to generate a 2-D color contour plot, which shows the location of tumor. Parametric analysis is carried out to fix the slot dimension to get optimum antenna performance. After successful simulation, the antenna structure is fabricated, and testing is carried out using Power Network Analyzer.

Constructing and utilizing toroidal modes in clusters of dielectric particles opens pathways to creating more efficient, compact, and functional devices across various fields, from sensing and telecommunications to energy and defense applications. Toroidal modes contribute to unusual material properties related to artificial magnetism, which is essential for designing innovative metamaterials. In this paper, we establish a relationship between eigenoscillations (modes) and scattering characteristics of a toroidal dielectric particle (torus) and clusters of particles composed of different numbers of dielectric disks arranged in a circular configuration (rings) in terms of the manifestation of their toroidal response. In particular, we examine the multipole contributions to the scattering cross-sections obtained in the exact form and long-wavelength approximation. A toroidal mode is introduced as a mode of the system for which the second-order term related to the exact electric dipole in the multipole decomposition is much greater than the first-order term. We show that the individual modes of the torus and hybrid modes of the ring consisting of an electromagnetically coupled ensemble of particles can be uniquely related, including the lowest-frequency toroidal dipole mode. Unlike the torus, the toroidal dipole mode in the ring can be separated in frequency from other multipole contributions, allowing excitation of the pure toroidal dipole resonance when providing corresponding irradiation conditions for external electromagnetic waves. This study provides an opportunity to better understand the physics of toroidal resonances in structures containing ensembles of dielectric particles and the peculiarities of their application in advanced microwave and photonic systems.

By using the time-dependent three-dimensional coupled-wave theory (3D-CWT), the transient analysis of photonic-crystal surface-emitting lasers (PCSELs) with double-lattice photonic crystals is performed. By optimizing the size of the PCSELs and the shape of the double-lattice photonic crystals, the resonance frequency is increased, and the damping (photon lifetime) is decreased, which enables over 40 GHz intrinsic 3 dB modulation bandwidth of the PCSELs. 100 Gb/s open eye under non-return-to-zero (NRZ) modulation is demonstrated by using such PCSELs. The large bandwidth enables single-lane 200 G optical transmission under four-level pulse-amplitude modulation (PAM-4). This study shows the design principles of large-bandwidth PCSELs and promises PCSELs to be an ideal candidate for the application of high-speed, high-power, free-space optical communication.

This research paper presents a novel dual-band slot array cavity-backed metamaterial antenna designed using advanced substrate integrated waveguide (SIW) technology. The antenna is specifically optimized for operation in the Ka-band frequencies. The incorporation of slots in the proposed design yields substantial advantages, such as enhanced impedance matching, exceptional directivity, and the capability for dual-band operation. Furthermore, the designed antenna showcases a gain of 6.40 dBi at 27.42 GHz and 4.77 dBi at 28.70 GHz, which is a result of the innovative incorporation of metamaterials within the SIW cavity. This demonstrates the antenna's ability to provide efficient signal reception and transmission, particularly in the specified frequency bands. The antenna's capability to operate in dual bands and its exceptional performance have been confirmed through rigorous simulation and experimental validation. These results substantiate its reliability and suitability for demanding millimeter-wave applications. The compact design and exceptional performance metrics of the proposed antenna underscore its potential for seamless integration into contemporary wireless communication systems, thereby laying the groundwork for continued progress in SIW and metamaterial-based antenna technology.

This paper presents a compact single-feed, rectangular slotted-patched antenna (SPA) for UWB applications. The proposed design adds a triangular part of the tail of the rectangular patch, cuts the edge of the patch, etches a rectangular slot in the ground plane, and then tunes the basic parameters of the design to achieve the UWB passband. The proposed antenna including slots on the patch for compact functionality is readily identifiable. The bandwidth and realized gain of the UWB antenna can be extremely improved to show the ability of a slot loading technique. The new conception of the rectangular patch antenna is considered. A feed mechanism using an inset patch feedline is implemented and analyzed. The parameters of the antenna are demonstrated, and the antenna is fabricated with an inexpensive FR4 substrate and validated experimentally. The antenna occupies frequency band (2.56-12) GHz. Making slots in the modified patch results in a significant gain improvement of 4.8 dBi as well as extending the UWB passband. The measured values of the reflection coefficient, VSWR, realized gain, and power pattern are in good agreement with the simulated results.

A band-pass filter utilizing a dual-mode Substrate Integrated Waveguide (SIW) cavity, enhanced by a novel Defected Ground Structure (DGS) is proposed in this paper. The SIW cavity operates in TE₁₁₀ and TE₁₂₀ modes, and the electric field of TE₁₁₀ is modified by introducing a series of metallized disturbance holes at the center of SIW cavity to increase the resonant frequency of TE₁₁₀ mode to that of TE₁₂₀ mode, thereby forming a passband with two transmission poles. A DGS that combines a dumbbell structure with a Complementary Split Ring Resonator (CSRR) is employed on the ground plane of the filter to improve the stopband rejection and suppress the parasitic passband. EM simulation and measurement results suggest that the center frequency of the filter is 4.8 GHz. It achieves a 3 dB-bandwidth of 300 MHz, with its insertion loss in the passband up to 0.5 dB and return loss greater than 20 dB. The designed DGS introduces a transmission zero near 7.2 GHz to suppress the parasitic passband and enhance the selectivity of the filter, while maintaining the original insertion loss and return loss within the passband. Its overall layout is simple and innovative. The designed filter is specifically engineered for application in the receiver of Nonlinear Junction Detection (NLJD) systems, aiming to suppress interference signals and allow only the second harmonic to pass through, which holds certain practical significance in RF engineering.

A single-band, linearly polarized Substrate Integrated Waveguide (SIW) antenna is designed specifically for WLAN 802.11a applications. The SIW design consists of four rectangular slots adjacent to each other through the SIW wall, with appropriate rectangular patch elements inserted in the two vertical slots for bandwidth enhancement. The structure is optimized to radiate at a frequency of 5.22 GHz, resulting in linear polarization caused by the excitation of the TE110 mode. The simulated design offers a gain of 7.275 dBi and a bandwidth of 47 MHz. The radiation pattern of the proposed fabricated antenna is measured in test environments where it is found to be unidirectional. The proposed design is compact and minimal in complexity, offering a higher gain.

This paper presents a dual-band microstrip filter with left-handed characteristics, featuring high selectivity and miniaturization. The design achieves negative permittivity and permeability by integrating H-shaped complementary split-ring resonators (CSRRs) within a substrate integrated waveguide (SIW). To enhance out-of-band rejection performance, a defected ground structure (DGS) is introduced. By applying the Half Mode SIW (HMSIW) principle, the equivalent magnetic walls of the SIW are cut, resulting in a 50% size reduction. Dual-frequency characteristics are realized using a symmetrical H-shaped CSRR, with the filter operating in the Sub-6G frequency band. Experimental results demonstrate that the filter exhibits good selectivity and low insertion loss at 3.5 GHz and 4.8 GHz. Tuning of the second frequency band is achieved by adjusting the coupling distance between the CSRR and metal via. This work has significant application potential in the fields of wireless communication and RF technology. The study provides theoretical support and technical insights for the design of future compact multi-band filters.

This paper investigates the effects of winding configurations on force density and fault tolerance in linear primary permanent magnet vernier (LPPMV) machines. Firstly, the LPPMV machines with integral slot distributed windings (ISDWs) and fractional slot concentrated windings (FSCWs) are discussed. Due to the high modulation ratio of ISDW machine, it has the potential to achieve higher thrust force capabilities. Then, the operation principle of the LPPMV machines is analyzed from the perspective of air-gap magnetic modulation. Furthermore, it should be noted that the winding configurations of ISDW machine has larger spans, resulting in insufficient fault-tolerance. To solve this limitation, a new modular ISDW LPPMV machine was proposed and optimized. In the modular ISDW LPPMV machine, a 3×3-phase winding configuration is employed. It is demonstrated that modular ISDW LPPMV machines exhibit superior characteristics in both thrust density and fault tolerance. Finally, the experiments are carried out in a linear test bench, verifying the theoretical analysis.

This paper introduces an artificial intelligence (AI) methodology designed to enhance the output of two-dimensional (2D) electromagnetic imaging systems, specifically tailored for the imaging of conductive objects utilizing inductive sensors. The core of our imaging system comprises a commercial data acquisition board, alongside custom-made multilayer planar coils developed by conventional printed circuit board technology. By leveraging recent advances in AI and machine learning, our approach significantly improves the resolution and clarity of electromagnetic images. The paper uses a multi-layer perceptron (MLP) classifier to process the raw electromagnetic data captured by the imaging system. These algorithms are trained to recognize patterns and anomalies in electromagnetic field data, which are often indicative of conductive objects. The enhanced imaging capability is demonstrated through a series of experiments that compare the AI-enhanced outputs with the ground truth.

Power amplifiers in wireless communication systems can introduce nonlinear distortion, degrade signal transmission quality, and increase power consumption. The paper presents a BiLSTM-CNN-based model for modelling power amplifier behaviour to address this issue. The model uses BiLSTM layers to capture temporal information from the signal data and incorporates a multi-head attention mechanism to focus on different temporal features. Additionally, convolutional layers process global features and reduce model parameters through weight sharing. Using this model, a digital pre-distortion (DPD) model is proposed to linearise the power amplifier through an indirect learning approach. The results show that the BiLSTM-CNN model achieves a normalised mean square error (NMSE) of -40.3dB, and the DPD model enhances the adjacent channel power ratio (ACPR) of the communication system by 18dB, demonstrating the model's feasibility. Comparative analysis with other network models indicates that BiLSTM-CNN outperforms traditional methods of fitting performance and convergence speed, showcasing its superiority.

This work derives exact expressions for the voltage and current induced into a two conductors non-isolated transmission line by an incident plane wave. The methodology is to use the transmission line radiating properties to derive scattering matrices and make use of reciprocity to derive the response to the incident wave. This methodology to derive receiving characteristics from the radiation properties via a scattering matrix is novel, and we already started to implement it to additional cases. An immediate advantage we obtained from this method is the derivation of a very simple analytic expression for the voltage and current for a matched transmission line. The analysis is in the frequency domain, and it considers transmission lines of any small electric cross section, incident by a plane wave from any direction and polarization. The analytic results are validated by successful comparison with ANSYS commercial software simulation results, and compatible with other published results.

For the purpose of fifth-generation and beyond communication applications, broadband circularly polarized (CP) & high gain AI-tuned metantenna operating in the 5 GHz band is presented in this article. So, an linearly polarized (LP) printed monopole antenna is being taken into consideration in the initial stage. To initiate CP from LP, a metallic strip that functions as a dynamic switching mechanism is utilized to short one of the parasitic conducting strips (PCS) with partial ground plane. The objective is to enhance the impedance (IBW) and axial bandwidths (ARBWs) as well as the antenna gain in order to make it a suitable candidate for ambient RF energy harvesting/wireless energy harvesting application. To achieve this, AI-tuned metasurface is placed below the monopole radiator at a height of 0.33λ_o. With a measured 49.84% IBW, 22.36% ARBW, CP gain > 8 dBic, antenna efficiency > 70%, fabricated on an FR-4 substrate with 1.3λ_o x 0.9λ_o x 0.02λ_o, it is suitable for the technological deployments in a current wireless technology, assuring resilience in networks. To meet the ever-increasing requirements of the current scenario, wireless communication landscape is on a paradigm shift. This transformation is brought by the utilization of metasurfaces offering customized, effective, and typical control of electromagnetic waves keeping with the desired frequency conditions.

Optimal placement of wireless base stations in urban areas allows for maximum coverage and performance whilst maintaining minimal cost. In this paper, we propose a novel machine learning approach to place base stations rapidly in an urban environment for 5G communications and beyond. This is a noteworthy approach as 5G, especially those that involve millimeter wave frequencies tend to require significantly higher number of base stations for any particular area, unlike their counterpart low frequencies where a small number of base station is sufficient to cover a good geographical area. Our machine learning empowered path loss model is developed to tackle this change in gameplay head-on, and it bridges the gap between empirical and ray tracing methods where we achieve accuracy closer to ray tracing yet at a significantly lower computation cost. Promising preliminary results are obtained, with a minimum coverage area of 80% with potential for future improvements.

This paper presents a coaxially fed, miniaturized tri-band circularly polarized (CP) antenna with a single layer patch configuration. Broadside radiation is achieved in the L5 band (1.176 GHz), L1 band (1.575 GHz), and S band (2.49 GHz), through the strategic excitation of higher-order modes (TM20 and TM30). The antenna design integrates slots and capacitors to reduce the operating frequency, efficiently excite all three modes, and achieve circular polarization within the designated bands. Each frequency band can be independently tuned with minimal effect on the performance of other bands. Moreover, it also facilitates the tuning of polarization sense (from RHCP to LHCP and vice versa) across all three bands. The proposed antenna radiates RHCP at L5, L1, and S bands, with a gain of 1.3 dBi, 1.5 dBi, and 2.7 dBi, respectively. A prototype with dimensions of (0.15λ_L5 × 0.15λ_L5) has been developed and fabricated to validate the antenna's performance.