Radar-based hand gestures recognition have played an important role in developing human-computer interaction (HCI). However, when radar-based hand gesture recognition techniques are applied to multi-target scenarios, the challenges mainly involve problems of mutual interference and inaccurate recognition of hand motion when both hands move within the same plane. Here, we propose a dual-hand trajectory perception prototype based on a 60 GHz frequency-modulated continuous-wave (FMCW) multiple-input multiple-output (MIMO) radar sensor platform with an L-shaped virtual antenna array. To address the challenges, the approach involves estimating azimuth and elevation angles separately from the two data components derived from the L-shaped array through multiple signal classification (MUSIC) algorithm, incorporating spatial division techniques combined with digital beam formation (DBF). The dual-hand applications mainly include angle targets at two distinct distances or dual-angle targets at the same distance. Therefore, the target distances are first determined using range fast fourier transform (range FFT). If a single target distance is identified, we proceed to solve for the angles of two targets. Alternatively, if two distinct target distances are distinguished, we individually solve for the single-angle target corresponding to each distance. Furthermore, to mitigate noise inherent in the raw data of visualization, a frame point removal and smoothing algorithm is devised to refine the trajectories. Experimental verifications prove that the proposed multi-target motion perception algorithm by using a MIMO FMCW radar sensor platform can realize accurate recognition of air-writing gestures and enable tracking the trajectories of both single-handed and dual-handed targets in three-dimensional space. It also gives a new option for controlling the HCI.

This article presents a compact, triple-band half mode substrate integrated waveguide (HMSIW) based self-multiplexing antenna (SMA) designed for various X-/Ku- band applications. The proposed SMA comprises a compact HMSIW with slots of unequal widths excited with three different ports resembling the anatomy of an arrow. These three slots are driven by a 50 Ω microstrip line feed, facilitating radiations at 10.82, 12.28, and 13.95 GHz with good isolation between the ports. Independent functioning at three different frequency bands is made possible by the remarkable versatility of the proposed SMA design method. With the isolation of nearly 20 dB between ports and gains of 3.97, 4.62, and 7.55 dBi at ports resonating at three distinct frequencies, the SMA-HMSIW element's total arrangement allows for a small antenna size of 0.44λ_g² at the lowest frequency of operation. The proposed SMA structure has been fabricated, and the results are measured which show good agreement with simulated ones.

This paper proposes a four-port ultra-wideband (UWB) MIMO antenna with high isolation and three notch bands. A cross-shaped decoupling structure (CSDS) is synergistically integrated with an improved L-shaped ground plane to achieve high port isolation. Composite resonant slots (CRSs) are introduced on the radiating patch, and two L-shaped slots are etched on the feed line to suppress interferences from WiMAX (3.19 GHz), C-band (4.45 GHz), and X-band (7.95 GHz). Simulation and measurement results verify that the Voltage Standing Wave Ratio (VSWR) at the center frequencies of the notched bands is greater than 6.8, 8.1, and 4.3, respectively. In the operating frequency band of 2.7-12 GHz (excluding the notched bands), the isolation is > 24.3 dB, envelope correlation coefficient (ECC) < 0.0045, diversity gain (DG) > 9.9990 dB, total active reflection coefficient (TARC) < -10 dB, and channel capacity loss (CCL) < 0.4 bps/Hz. It fully meets the requirements of high-performance MIMO systems for channel independence and transmission efficiency. Compared with similar studies, this work has significant advantages in core indicators such as bandwidth, number of notched bands, and isolation, providing new ideas for the design of UWB-MIMO systems in complex electromagnetic environments.

Unmanned aerial FPV systems demand ultra-low latency, high-reliability communication. At high speeds and in cluttered environments, Doppler shifts and rapid multipath changes significantly increase packet error rates (PER). This paper introduces a novel solution: real-time geometry tuning of a circularly polarized helical antenna array to mitigate these effects in ExpressLRS (ELRS) long-range FPV control links. Using full-wave simulations (Ansys HFSS) and blind field trials, we validate system performance. A new analytical framework integrates Doppler-induced frequency offset into the antenna's radiation pattern and PER model. The adaptive array autonomously adjusts coil pitch and diameter based on velocity and attitude, reducing PER by 20% at speeds over 150 mph. It also maintains near-unity VSWR, preventing reflection spikes, and halves RSSI variation, indicating improved link stability. These results demonstrate that tunable helical antennas can effectively mitigate Doppler and multipath impairments in high-mobility UAV environments, informing future antenna designs and supporting the development of AI-integrated, adaptive RF systems for drone racing and autonomous swarms.

This work explores the data-driven approaches for breast tumor detection and analysis of different breast tissues by using microwave sensing technique. Microwave sensing offers a promising trade-off in tissue penetration depth and is prominent dielectric disparity between healthy and tumorous tissues at microwave frequencies. Tumor cells exhibit unique properties, such as increased water content and different ionic composition, which create distinct dielectric traits compared to healthy tissue. This frequently shows variations in loss characteristics compared to normal tissue and can exploit those differences for detection. The key parameter used is Specific Absorption Rate for the determination of tumor location. The differential absorption between healthy and tumor tissue is potentially aided in identifying the presence of lesion. The five sets of reflection characteristics are recorded with the system comprising UWB antenna with breast phantom by using VNA with a gap of 4-5 days. Further, the dimensionality reduction technique is applied to extract the features using PCA and tSNE. In order to enhance the detection accuracy, dimensionality reduction techniques are used in tandem with the supervised machine learning approach. Among the four supervised algorithms, including SVM, KNN, RF and MLP, the random forest was found to be the most optimal for the data with an auc score of 99.97%.

Nowadays, microelectronic integrated circuit (IC) design constitutes the biggest challenge of negative group delay (NGD) electronic engineering research. Bandpass (BP) type NGD circuits are generally designed with resonant and not-integrable inductive large size network-based topology. However, BP-NGD circuit integrability is delimited by the inductor design. A design solution for fully resistive-capacitive (RC) network-based BP-type IC in 130-nm CMOS technology is the purpose of the present research work. The theory expressing the design equations of RC-network based BP-NGD circuit is developed. The design feasibility is verified with a proof-of-concept (POC) represented by a 130-nm CMOS RC-network passive IC with 0.68 mm × 0.72 mm physical size simulated by Cadence®. The obtained results of S-parameters confirm the BP-NGD behavior of the CMOS IC POC with 21.9-MHz NGD center frequency and -0.99-ns NGD value over 68-MHz NGD bandwidth. The BP-NGD characterization results are in excellent agreement with the theoretical model. The robustness of 130-nm CMOS BP-NGD RC passive IC is explored by 2000 trials Monte Carlo statistical analysis with respect to the uncertainty of component parameters.

Flux-reversing machines (FRMs) have the advantages of high torque density and wide speed range. However, their disadvantage is the low utilization rate of permanent magnets (PMs). To enhance PM utilization, a novel dual-PM FRM (DPFRM) with Halbach arrays is proposed in this paper. Halbach arrays are applied to both the stator interlayer and half of the rotor teeth, forming a consequent-pole structure together with iron cores. This layout significantly reduces the number of rotor magnets used. With the use of Halbach arrays, this design effectively reduces magnetic flux leakage. It also achieves higher torque density under low current conditions, demonstrating enhanced electromagnetic performance. To achieve better overall performance, both the conventional FRM and the proposed DPFRM are globally optimized. Their no-load and load performances are evaluated through finite element analysis (FEA). The analysis verifies that the DPFRM achieves higher back electromotive force (Back-EMF) and torque density, and also exhibits lower torque ripple. Therefore, the proposed design significantly improves PM utilization, effectively mitigating the primary limitation of conventional FRMs.

This paper presents a miniaturized broadband port-reuse microstrip circuit to address the challenges of bulky volume, excessive insertion loss, and parameter deviation superposition caused by discrete port design and discrete circuit design in the interconnection between active phased array antennas and T/R components. Based on an integrated design methodology, the circuit achieves bandpass filtering, bidirectional power coupling output, DC power supply port functionality, and RF/DC isolation through a single-port interconnection. Experimental results demonstrate that the implemented circuit in Ku-band exhibits 13.5-15.18 GHz bandpass filtering characteristics, bidirectional signal power monitoring capability, 0-12 V/2.5 A DC power supply functionality, and effective RF/DC signal isolation. The measured results align well with theoretical predictions. This architecture demonstrates exceptional adaptability and seamless integration capability, showing significant potential for large-scale deployment in various transceiver architectures such as satellite communication systems.

To improve the current and torque regulation performance of the traditional deadbeat predictive current control (DPCC) for switched reluctance motors under model parameter mismatch, this article proposes an improved DPCC method based on the sliding mode strategy. First, a dedicated torque-current converter is formulated to achieve precise transformation of electromagnetic torque into corresponding q-axis current references. Second, a unified anti-disturbance sliding mode control compensation scheme is introduced into both the torque-current converter and the deadbeat controller to mitigate the negative effects of model parameter mismatch on current and torque control. This integration achieves indirect torque control through phase current modulation, effectively reducing torque ripple. Furthermore, the stability of the controller under model parameter mismatch conditions is rigorously demonstrated through Lyapunov stability analysis. Finally, the effectiveness of the proposed control method is demonstrated through simulation results, and its significant superiority in current control performance and torque ripple suppression is shown.

This paper presents the design, simulation, and experimental validation of a 1-bit reconfigurable reflectarray. Each unit cell is equipped with a single PIN diode, enabling binary phase control (0˚ and 180˚) for dynamic beam steering. The reflectarray employs a compact and cost-effective architecture, with digitally reconfigurable elements that allow real-time control of the reflected wavefront. The integration of PIN diodes ensures fast switching and low power consumption while maintaining good reflection efficiency and phase performance. Without limiting the applicability of the method, a reflectarray antenna consisting of 9×9 element array operating at 5.8 GHz was designed. Full-wave electromagnetic simulations and measurements demonstrate beam steering capability up to ±15˚, with minimal gain degradation and acceptable side-lobe levels. The proposed reflectarray design is a promising solution for low-complexity, reconfigurable antenna systems in applications such as wireless communications, radar, and IoT systems operating in the 5.8 GHz ISM band.

This paper presents the development of a low-frequency dual-port microwave sensor designed for permittivity detection in both solid and biological materials. The sensor integrates a circular split-ring resonator (CSRR) with an electric field coupled (ELC) structure on a planar dielectric substrate, resulting in a compact and simple architecture that supports ease of fabrication and low-cost implementation. Operating at a resonant frequency of 0.86 GHz, the sensor is particularly suitable for characterising biological samples such as meat, fish, squid, and chicken, as lower frequencies offer deeper penetration and better interaction with high-loss biological tissues. Validation through full-wave simulation and experimental measurement confirms the sensor's capability to detect permittivity variations across a wide range of materials. A polynomial fitting model is employed to extract permittivity values based on resonance frequency shifts, achieving accurate results with a maximum error below 7% and overall accuracy exceeding 93%. The device demonstrates reliable performance in estimating permittivity values from ε_r = 1-9.8, including unknown biological samples with normalized sensitivity of 0.02% and frequency detection resolution 0.019 GHz. Measurements show clear frequency shifts that correlate with dielectric changes, and the experimental results align closely with the simulation data. The simple structure of the sensor also supports straightforward integration with common measurement instruments such as vector network analysers, making it practical for real-time monitoring and portable applications. The low operating frequency combined with the straightforward design provides an effective solution for applications requiring permittivity detection of lossy, heterogeneous, or biological materials. This work contributes a feasible and efficient sensor platform for use in medical diagnostics, food quality inspection, and other industrial contexts where reliable, low-cost dielectric sensing is essential.

In this paper, a compact microstrip line (MSL)-to-circular waveguide (CWG) transition using a rectangular patch is introduced. As the rectangular patch is placed 2.68 mm (0.043 λg) away from the short-circuited plane of the CWG, the transition is very compact. By truncating the rectangular patch of the compact MSL-to-CWG transition, a compact MSL-fed CWG polarizer using the corner-truncated patch is proposed. The proposed polarizer has a compact size and a phase difference of -90.97° at 9.65 GHz. The axial ratio is within ±1 dB from 8.5 GHz to 10 GHz. The reflection coefficient is smaller than -10 dB from 9.03 GHz to 10.5 GHz. In addition, as the corner-truncated patch is adopted, the proposed polarizer does not require a complex manufacturing process on the waveguide. Moreover, as the microstrip line feeds the polarizer, it can be easily integrated with other planar circuits. To verify the simulation results, the MSL-fed CWG polarizer using the corner-truncated patch is fabricated and measured. The simulation and measurement results are in good agreement.

This article presents a compact and flexible ultra-wideband (UWB) antenna with a defected ground structure (DGS) for IoT applications. The antenna is fabricated on a 0.7 mm thick jeans substrate to ensure high flexibility and take advantage of its universal availability. Machine learning techniques are applied to optimize the antenna's performance. A ring-shaped patch with DGS and C-type stubs is used to achieve a large bandwidth and reduce size. The total dimension of the proposed flexible antenna is 38 × 26 × 1.7 mm³. The primary aim of this article is to design a flexible UWB antenna with a remarkable impedance bandwidth of 131.45%, which covers frequencies from 2.56 GHz to 12.38 GHz. It operates at 3, 8, and 11.32 GHz frequencies with 99%, 98.30%, and 96.71% radiation efficiencies, respectively. The realized gain is 2.51, 3.70, and 5.46 dBi at frequencies 3, 8, and 11.32 GHz, respectively, with a peak gain of 5.46 dBi at 11.32 GHz. Specific absorption rate (SAR) values were tested using a human phantom and met FCC limits, confirming suitability for wearable and flexible IoT applications. The design was optimized using machine learning (ML), with KNN performing best, achieving 99.5% accuracy in S-parameter prediction. The measured and simulated results are correlated with each other for flat and bent antennas.

Microwave-based object localization system is a noninvasive technique that uses microwave signals to detect, map, and analyze the properties of materials. This approach provides information about hidden objects within materials. However, the localization process can be complex, requiring sophisticated algorithms to interpret the signals accurately. This study proposes a new technique for microwave-based object localization system using Radio Frequency (RF) Codes to perform spatial detection with four pairs of RF Code sensors representing bits of ``111,'' ``110,'' ``101,'' and ``011.'' The system incorporates four identical RF Code paths arranged symmetrically around a circular container, improving spatial coverage and enabling accurate detection of hidden objects located at eight different spatial positions. Steel is used as the hidden object, while Stone serves as Material X in this system. The system achieved an average detection accuracy of 70% and a detection efficiency close to 100% across all spatial positions. Additionally, the RF Code performance chart is designed to interpret the detection accuracy results, making the analysis more accessible and practical. The proposed system has potential applications in nondestructive testing, material analysis, industrial inspection, and security systems, offering a reliable and efficient solution for detecting hidden or embedded objects.

This paper examines the impact of soil stratification and the frequency dependence (FD) of the Earth's electrical parameters on the transient response of underground cable systems, accounting for both earth-return admittance and impedance. A derivative-free iterative approach is proposed to overcome issues of discontinuous modal transformation matrices that occur at certain frequencies when using conventional diagonalization algorithms. This method ensures smooth and continuous eigenvector tracking. Transient voltages and currents along cables are computed using a modal-domain-based transmission line model combined with Numerical Inverse Laplace Transform (NILT). Simulation results validate the proposed method's accuracy and stability, and highlight the significant influence of the stratified frequency-dependent (SFD) ground under various operating conditions. Finally, a reduced equivalent model of the three-phase underground system is established to facilitate further analysis.

This paper presents a CPW-fed Minkowski island fractal monopole antenna with wideband characteristics. The Minkowski island fractal geometry is applied on the radiating patch of the monopole antenna to make it compact and enhance bandwidth performance. The measured return loss values indicate a fractional bandwidth of 114% from 2 GHz to 7.3 GHz. The simple structure and wideband characteristics make this antenna suitable for various wireless communication applications, including WLAN, Wi-Fi, Wi-Max, 5G, and sub-6 GHz band services.

With the development of deep learning networks, convolutional neural network (CNN) and other techniques provide effective detection methods for synthetic aperture radar automatic target recognition (SAR ATR), and have been widely used. However, due to the objective factors such as complex scene interference inherent in SAR images, the recognition rate of traditional time-domain processing of SAR images is not high enough, which is still a key problem to be solved urgently. To solve this problem, we propose a space-frequency dynamic fusion network (SDF-Net). The network consists of four space-frequency joint processing (SJP) modules connected in series, each comprising convolutional layers and unbiased fast fourier convolution (UFFC) units at different scales to achieve joint feature extraction in the spatial and frequency domains. Building on a four-level series structure, residual paths from the original image features are introduced into the inputs of SJP2, SJP3, and SJP4. Additionally, residual paths from the features output by SJP1 are introduced into the inputs of SJP3 and SJP4, and from SJP2 into the input of SJP4. By incorporating residual paths of features from different stages, the network facilitates cross-stage information interaction, effectively integrating long-distance contextual information. At each fusion node, dynamically generated weights are used for feature fusion, followed by sequential progressive processing through spatial-frequency joint processing, ultimately leading to classification and recognition results. Experimental results on the MSTAR dataset and the FUSAR-Ship1.0 dataset show that compared to traditional methods, this network algorithm achieves a higher recognition rate.

In recent years, harvesting abundant, clean and renewable wave energy from the ocean has become one of the most promising ways to obtain electricity. However, the multi-directional nature of waves and the low frequency of the movement pose a current challenge. We designed and fabricated an all-directional triboelectric nanogenerator (AD-TENG) with a biomimetic snail golden spiral. It consists of an electric energy collector in the vertical direction and another in the horizontal direction, and mainly operates through the contact separation method. The AD-TENG converted mechanical energy into electrical energy with the swinging of the pendulum and folding movements. The spiral structure of AD-TENG can harvest energy in all horizontal directions, and the wiring is simple, requiring only two positive and negative wires. In a water wave environment, the AD-TENG charged a capacitor of 100 μF to a voltage of 3 V in 2 min, lighting up 150 LED bulbs. The experiment measured the peak-to-peak voltage (Vpp) from 15 different angles and calculated the error ``e'' as 4.6%. The multi-degree-of-freedom energy harvesting and adaptability to various water wave motions of the AD-TENG offer great potential for the development of self-powered marine sensors.

This paper presents a comparative study evaluating the influence of monostatic and multistatic microwave imaging (MWI) configurations on imaging performance. Localization accuracy and Signal-to-Noise Ratio (SNR) are evaluated as key performance metrics for both configurations. Numerical simulations are conducted using CST Studio Suite, considering various scenarios involving circular antenna arrays surrounding embedded metallic rebars of different sizes within concrete pillars of varying geometries. Image reconstruction is performed using the Delay-and-Sum Integration (DASI) algorithm, an enhanced version of the conventional Delay-and-Sum (DAS) technique. The simulation results show the performance of the proposed reconstruction technique in terms of localization accuracy.

A multiband planar antenna fed by an asymmetric coplanar waveguide (ACPW) is proposed and fabricated. The design incorporates branched stubs within the split-ring resonator (SRR) and integrates this modified SRR with the ACPW structure, thereby expanding the antenna's operational bandwidth and improving gain performance. The antenna has dimensions of 60 mm × 58 mm × 0.813 mm, which are equivalent to 0.48λ₀ × 0.46λ₀ × 0.0065λ₀ at 2.4 GHz. Simulation and measurement results demonstrate close agreement. The antenna exhibits |S₁₁| < -10 dB in the frequency bands of 2.23-2.51 GHz, 3.54-4.47 GHz, and 5.01-6.29 GHz, with a maximum gain of 7.07 dBi at 5.1 GHz and over 2.5 dBi gain across all bands. This antenna meets the requirements for WLAN, UAV communications, and 5G applications.