This letter presents a hybrid-controlled multifunctional metamaterial (HCMM) that integrates electromagnetic switching and frequency tuning through combined active and microfluidic modulation. The HCMM achieves rapid switching between shielding and transmission states via active control of p-i-n diodes, while wideband frequency tuning is accomplished by manipulating liquid metals (LMs) in embedded microfluidic channels. Wideband frequency adjustment is demonstrated through discrete coarse-tuning (LM injection/discharge) and continuous fine-tuning (precise LM segments repositioning). The underlying operational mechanisms are analyzed by establishing an equivalent circuit model. Finally, experimental measurements of a fabricated prototype validate the design, confirming its dual-function reconfigurability in both state switching and frequency adaptation.

The simulation of resonant structures poses a significant challenge for time-domain full-wave solvers due to their inherently slow energy dissipation, which would result in long-lasting electromagnetic (EM) oscillations. It often necessitates a prolonged simulation time to achieve numerical convergence or adequate signal attenuation for extracting the frequency-domain response. To address this issue, we first propose a robust derivative-enhanced generalized pencil-of-function (Der-GPOF) method integrated with the discontinuous Galerkin time-domain (DGTD) method. The Der-GPOF formulates an augmented matrix pencil by fusing the explicit time-derivative data naturally from the semi-discrete DGTD scheme with the second-order Leapfrog (LF2)-generated temporal signals, thereby embedding additional physical constraints. Second, the minimum description length (MDL) criterion facilitates the adaptive truncation of singular values for the pole extraction, effectively isolating physical resonances from noise without the need for an empirically defined threshold. As such, the proposed Der-GPOF method achieves superior stability and a shorter temporal sampling window than the conventional generalized pencil-of-function (GPOF) method. Finally, an adaptive algorithm is developed to obviate manual configuration of the temporal sampling window in the Der-GPOF method. By incrementally extending the sampling window duration and evaluating the reconstructed signal, the algorithm autonomously determines the truncation time of the DGTD simulation without any prior knowledge of resonant structures. In applications requiring frequency-domain analysis, the spectral response is analytically computed using the Fourier transform of the extracted pole-residue model. This analytical approach effectively overcomes the spectral leakage and resolution limitations inherent in the discrete Fourier transform (DFT). Numerical examples demonstrate that the proposed method achieves a remarkable gain in computational efficiency while maintaining high fidelity and accuracy in both the time and frequency domains.

Several cost-effective a posteriori error estimators applicable to the volume integral equation (VIE) are investigated, and their practical application scenarios are studied. These estimators utilize discontinuities in physical quantities or recovered fields to perform error estimation. Despite their low computational cost, they demonstrate excellent ability to identify regions of significant error. The global prediction error is defined by aggregating the local errors computed from each element using these estimators, and it serves as a termination criterion for adaptive refinement. To assess the accuracy and reliability of the global prediction error, we analyze its Pearson correlation with two metrics: the root-mean-square error (RMSE) of the radar cross section (RCS) and the reference error. Numerical results confirm a strong correlation between the global prediction error and the RMSE as well as the reference error, while different error estimators show varying degrees of correlation. This article presents an adaptive $h$ -refinement algorithm for the VIE, based on cost-effective a posteriori error estimators. This algorithm accurately identifies regions with significant errors within the mesh and adaptively refines the corresponding tetrahedra. Numerical results show that the adaptive refinement achieves significantly higher convergence rates for both the RMSE and the reference error than those achieved by global refinement. Furthermore, by strategically allocating computational resources to critical regions, the proposed algorithm achieves an optimal balance between accuracy and computational efficiency for complex electromagnetic problems with significantly fewer unknowns.

Numerical Green's functions (NGFs) encapsulate the effects of complex media as precalculated functions, significantly reducing the number of unknowns in electromagnetic scattering problems. However, recalculations are often required when scatterers' positions change. In this article, a neural network-accelerated NGF method is proposed for 3-D randomly distributed dielectric targets. This method enables real-time NGF output of targets at arbitrary positions within the domain. By analyzing three cases-a single dielectric target, a group with identical targets, and a group with different targets-the field and source, along with the spatial distribution and properties of scatterers, are preprocessed as input of the Kolmogorov-Arnold network (KAN). The incorporation of physical information allows the network to effectively capture the impact of scatterers' position variations on the NGF. In addition, feature extraction is employed to enhance the efficiency and reduce the mean squared error (mse) of KAN by extracting the unknown NGF secondary term as the acceleration part. The proposed neural network-accelerated NGF method is applied to the analysis of 3-D metamaterial scattering and the approximate modeling of indoor wireless channels. Numerical results demonstrate that the trained network model can output the NGF for arbitrarily distributed targets, significantly reducing the computational cost of scattered field calculations while maintaining low numerical errors.

This letter presents a novel radar cross section (RCS) enhancement structure: a dual R-KR lens retrodirective array (LRDA). A single R-KR LRDA is first designed by utilizing the broadband and wide-angle beam scanning capabilities of the R-KR lens. However, this configuration exhibits significant monostatic RCS fluctuations across the angular range, as observed in the simulation results. An analysis of the field distribution identifies the primary causes of these fluctuations. To address this issue, a dual R-KR LRDA is proposed, in which two single R-KR LRDAs are arranged in a parallel, laterally staggered configuration. Simulation and measurement results indicate that the dual R-KR LRDA achieves a maximum monostatic RCS fluctuation of approximately 3 dB over an angular range of +/- 55 degrees within the frequency range of 8.1 GHz to 12 GHz under linearly polarized incidence, demonstrating enhanced retrodirective performance.

Pattern synthesis for large-scale spherical array antennas (SAAs) has traditionally been computationally intensive. To enable flexible beam pattern control in SAAs, a real-time pattern synthesis method integrating the fast spherical Fourier transform (FSFT) with a lightweight deep neural network (LDNN) is proposed in this communication. The framework comprises three phases: sample database generation, neural network training, and online prediction. To efficiently collect training data, the direct and inverse Fourier transform relationship between the excitation distribution and the array pattern of a SAA is established using spherical harmonics, thereby extending the traditional fast Fourier transform (FFT) pattern synthesis algorithm, which is typically applied only to planar arrays, to SAAs. Numerical simulation results demonstrate that the proposed method based on the FSFT achieves significantly lower computational complexity compared with existing methods. In the LDNN training phase, an end-to-end model is introduced to approximate the input-output mapping of the FSFT-based method. An iterative trainingpruning mechanism is employed to minimize the redundancy of the neural network. Simulation results show that the obtained LDNN achieves a computational complexity reduction of over 50% in the online prediction process. For an SAA with 1196 elements, the focused beam patterns are synthesized in just 35 & micro;s

Orbital angular momentum (OAM) waves exhibit orthogonal phase characteristics and have attracted considerable interest in wireless communications, radar imaging, and electromagnetic sensing applications. This paper presents a compact planar OAM filtenna that integrates an incomplete uniform circular array (UCA), a spiral spoof surface plasmon polaritons (SSPPs) waveguide, and a half-mode substrate integrated plasmonic waveguide (HMSIPW) filter. The SSPPs feed excites sequentially rotated circularly polarized elements to generate vortex waves, while the HMSIPW introduces bandpass filtering without increasing the size. Theoretical analysis establishes the mode relation, revealing the OAM generation mechanism. Simulated and measured results confirm frequency-selective radiation from 4.12 to 5.82 GHz, with six modes and mode purities up to 0.96. The proposed filtenna achieves compact, high-purity, and frequency-reconfigurable OAM radiation, showing great potential for wireless communication and radar imaging systems.

Realistic vehicular channel attenuation traces are essential for attenuation-sensitive vehicle-to-everything (V2X) system design, simulation, and validation, yet large-scale acquisition of such measured traces remains costly and difficult across diverse road environments. To address this problem, this paper proposes a vehicular-channel-oriented generative adversarial framework, termed WGAN-GP-SAM, which incorporates self-attention and intra-batch mini-batch discrimination into a Wasserstein generative adversarial network with gradient penalty. Combined with an asymmetric update strategy and dynamic learning-rate scheduling, the proposed model is designed to improve long-range dependency modeling, enhance sample diversity, and alleviate mode collapse in channel attenuation sequence generation. The model is trained on real measurement data collected from eight vehicular scenarios. Comprehensive evaluations show that the generated sequences can reproduce the main statistical characteristics and temporal fading trends observed in the real measurements. In particular, besides similarity and coverage-imbalance comparisons and waveform inspection, second-order fading statistics, including level crossing rate (LCR) and average fade duration (AFD), are further examined, and a nearest-neighbor-distance-based analysis is conducted to check for possible memorization. To assess the engineering applicability and hardware reproducibility, channel replay experiments are implemented, where all tested scenarios satisfy the waveform fidelity requirement under the NRMSE criterion. Moreover, BER and BLER comparisons in representative scenarios show that the generated and replayed data preserve consistent SNR-dependent link-level trends and remain highly aligned with each other. These results suggest that the proposed framework provides a practical and better-validated pipeline for vehicular channel attenuation sequence generation and hardware-in-the-loop validation.

A new spectral integral method (SIM) based on divergence-conforming Gauss-Lobatto-Legendre (GLL) polynomials is proposed to solve the electromagnetic scattering problem for 3-D nonsmooth multilayered composite bodies of revolution (BoRs). The method is based on the mixed-order divergence conforming vector basis functions that have spectral accuracy. Unlike the conventional BoR-SIM with fast Fourier transform (FFT) acceleration, this new BoR-SIM can solve the problem of smooth as well as nonsmooth objects by incorporating the higher order basis functions and pole boundary conditions. Furthermore, the Poggio-Miller-Chang-Harrington-Wu-Tsai (PMCHWT) surface integral equation (SIE) is utilized alongside the method to avoid resonance difficulties when dealing with wideband scattering from multilayered BoRs with arbitrary shapes. Numerical examples are tested to verify and show the improvements in accuracy and efficiency. The results are compared with those of commercial software FEKO and the conventional BoR method. Its practical applicability is demonstrated by modeling multilayered objects with a perfect electric conductor (PEC) parabolic antenna in a radome. All numerical results show that the BoR-SIM is an efficient alternative to other BoR methods for scattering problems.

A novel approach is proposed in this letter, involving the integration of heterogeneous dummy elements surrounding the full array to realize scattering reduction. The utilization of heterogeneous dummy elements offers increased flexibility in quantity compared to homogeneous ones, resulting in a more significant reduction in monostatic radar cross section (RCS). Two planar phased arrays are developed to validate the proposed concept, including a 10 & times; 10 reference array and a 10 & times; 10 proposed array with 2 & times; 10 heterogeneous dummy elements. Both arrays exhibit scanning capabilities of up to 45 degrees in the E-/H-planes within the frequency range of 7.2 GHz to 10 GHz. The proposed array demonstrates a significant reduction in simulated monostatic RCS exceeding 20 dB across 7.2 GHz to 10 GHz when compared to an equal-sized metallic plate under linear-polarized normal incidence, with an averaged RCS reduction surpassing 24 dB. Prototype arrays are fabricated and measured to verify the proposed approach.

A low-sidelobe independently controllable multi- beamforming time-modulated array (TMA) with arbitrary stepped waveforms is proposed. Unlike conventional multibeam TMAs with constrained harmonic relationships, independent control of beam directions and amplitude weightings is achieved by assigning harmonic complex coefficients at different operating sidebands. The corresponding time-modulated waveforms are then obtained via inverse Fourier transform and quantized into practical stepped waveforms for implementation using in-phase/quadrature (I/Q) time modulators. Only a few global variables, whose number is independent of the array size, are optimized to improve efficiency while satisfying the sideband level (SBL) and sidelobe level (SLL) requirements. Experimental results from an eight-element prototype with far-field calibration verify independently controllable five-beam radiation with a measured SLL of −19.3 dB and an SBL below −30 dB.

This article introduces an inverse generative design system (IGDS) that leverages a physics-informed diffusion model for the interpretable design of frequency-selective surfaces (FSSs). Moving beyond conventional data-driven paradigms that often function as “opaque model,” the IGDS establishes a novel, three-way mapping between electromagnetic (EM) responses, equivalent LC circuit parameters, and geometric topologies. This framework enables three key breakthroughs: 1) topology innovation, generating unprecedented FSS structures that transcend the limitations of the training data; 2) intrinsic physical interpretability, achieved through a cross-attention mechanism that dynamically visualizes how metallic patterns are mapped to specific circuit parameters during generation; and 3) exceptional generalizability, facilitating the design for unseen performance metrics and scalable extension from single-layer to complex multilayer architectures. We validate this closed-loop framework—which seamlessly integrates specification interpretation, circuit synthesis, topology generation, and performance optimization—through rigorous FSS experiments. The results demonstrate that our system not only produces novel, manufacturable designs but also achieves a high degree of compliance with target specifications. This performance has been rigorously validated through both full-wave EM simulations and experimental measurements.

Using antenna arrays capable of generating arbitrary polarized electromagnetic (EM) waves has the potential to significantly enhance the adaptability of wireless electronic devices to increasingly complex EM environments. However, dynamically generating arbitrarily desired elliptical polarization (EP) radiation-controlling parameters, such as ellipticity, orientation, and handiness, with one antenna element per radio frequency (RF) channel remains a challenging task. In this work, we propose an arbitrary polarization wave radiator (APWR) enabled by a mechanism that configures the polarization states and excitation phases of elements in a multilinear polarization-reconfigurable (MLPR) antenna array. The proposed mechanism fully leverages the discrete polarization states of multiple array elements to synthesize polarization states that individual elements cannot directly generate. Combined with the excitation phase configuration, the desired EP beam can be directly generated under the premise of one element per RF channel. Equipped with devices for real-time control of the element polarization state and excitation phase, the proposed APWR can dynamically generate arbitrarily desired polarization radiation, ensuring effective polarization matching in complex environments. Through numerical examples and APWR prototype measurements, we validate the correctness and effectiveness of the proposed concept. The results suggest that this approach provides a promising pathway for integrating polarization as a dynamic resource in 5G, satellite communications, and the Internet of Things (IoT) networks.

Accurate simulation of multiscale electromagnetic structures with embedded fine features is often limited by the need for global remeshing and the difficulty of coupling nonconforming meshes. Classical embedded FEM–DDM formulations address this issue but rely on volumetric and port-based equivalent sources, which introduce analytical and computational complexity. This work proposes a simplified embedded FEM–DDM that removes all equivalent-source constructions. A single correction term, applied only within the overlapping region, accounts for material discontinuities and ensures consistent energy exchange between subdomains. The resulting scheme offers a significantly streamlined formulation while retaining the accuracy of conventional embedded FEM–DDM. Representative scattering examples, including radome–antenna configurations, demonstrate that the proposed method achieves comparable accuracy with reduced implementation effort and improved mesh reusability. The approach provides a practical and scalable tool for multiscale electromagnetic analysis.

A novel technique for the design of a three-dimensional multi-angle retrodirective metasurface (3D-MRDM) is put forward. Firstly, based on Floquet theory and the surface impedance theory, the surface current required for accurate retroreflection of the multi-angle retrodirective metasurface (MRDM) is deduced. Secondly, subcells of the MRDM are refined according to the surface current so that the MRDM can achieve high-efficiency retroreflection of multiple angles in the two-dimensional plane. The MRDM is finally extended to a 3D-MRDM by a specially designed honeycomb-meta (HM). A 3D-MRDM that can simultaneously achieve the efficient retro-reflection of a transverse electric (TE) polarized wave incident at seven angles on all three primary channels at every 120° azimuthal interval is designed as a conceptual example. Due to the unique configuration of the HM, the seven incident angles still have high retrodirectivity when the azimuth of the incident wave deviates from the three primary channels by ±5°. The measured results verify that the 3D-MRDM enables the retroreflection of seven incident angles simultaneously over an azimuthal range up to 30°. The design approach offers new perspectives for expanding the MRDM's angular domain, opening up opportunities for its application in radar electronic warfare as calibration targets or mounted on target drones.

This communication proposes a novel augmented electric field integral equation incorporating the impedance boundary condition (AEFIE-IBC). The proposed formulation eliminates spurious internal resonances while ensuring favorable matrix conditioning for the numerical simulation of fully or partially coated objects. Unlike traditional combined field integral equations (CFIEs)-which linearly combine the electric field integral equation (EFIE) and the magnetic field integral equation (MFIE)-the proposed AEFIE-IBC suppresses spurious resonances by explicitly enforcing the normal boundary condition. Importantly, it retains an electric field-based excitation on the right-hand side, thereby extending its applicability to both scattering and radiation problems. Furthermore, a Calder & oacute;n preconditioner based on a mixed discretization scheme is developed for the AEFIE-IBC, significantly enhancing matrix conditioning without compromising numerical accuracy.

This article proposes a novel iterative hp-adaptive refinement scheme for solving electromagnetic radiation and scattering problems through surface integral equations. The method introduces an improved current discontinuity error estimator that incorporates a weighting factor associated with the order of the basis functions defined on each element, enabling accurate error estimation within the hp-adaptive framework. To guide the refinement process, a control algorithm is developed that combines geometric model features with a solution variation indicator derived from the differential of current magnitude, which allows for a reliable and appropriate local choice between h- and p-refinement for each element. The refinement process starts from a coarse mesh and achieves high computational efficiency due to the lightweight implementation of both error estimation and control algorithms. In addition, a conformal mesh refinement strategy utilizing auxiliary meshes is introduced to maintain mesh quality across multiple refinement iterations. The proposed method is shown to be effective, accurate, and robust through a series of numerical examples.

This communication presents a multilevel reduced characteristic mode (CM) method that significantly enhances the efficiency of CM analysis for perfect electric conductor objects formulated with the electric-field integral equation (EFIE). This method utilizes CM-based basis functions (CMBFs) to substantially decrease the number of unknowns, successfully obtaining the eigenvalues with high modal significance (MS) and their associated mode currents. In addition, it constructs multilevel CMBFs by using the truncated mode currents and reducing the matrix size at each level to further reduce both computational costs and memory requirements. Our numerical experiments demonstrate that both the reduced CM equation method and its multilevel strategy can significantly reduce the number of unknowns by over 75% when applied to subdomains containing, on average, more than 300 Rao-Wilton-Glisson (RWG) basis functions. Although the algorithmic complexity remains unchanged, the proposed method achieves over 70% reduction in computational time compared with the traditional approach for models with fewer than 50 000 unknowns, while maintaining only minor degradation in the accuracy of mode currents.

To improve the computational efficiency of electromagnetic simulations, this article proposes a nonconforming finite element-boundary element (FEM-BEM) adaptive scheme. Unlike conventional approaches, this scheme leverages nonconforming flexibility to adjust local mesh density contrast between domains, while ensuring coordinated accuracy evolution through an adaptive process driven by domain-specific error indicators and a tailored bisection strategy. Specifically, a residual-based error indicator is formulated for the FEM domain, incorporating both interior discretization errors and mismatches with the BEM solution to drive the refinement process. For the BEM domain, a global residual-based error estimator incorporating the error contributions of surface equivalent electric and magnetic currents is employed to monitor convergence. Error indicators are introduced to quantify the discrepancies between the BEM and FEM solutions, referred to as coupling errors herein, which guide the local adaptive mesh refinement (AMR) of the BEM domain. A tailored bisection scheme ensures nesting between the BEM mesh and the FEM surface mesh. Numerical examples involving electromagnetic radiation and scattering demonstrate that, compared with conforming adaptive FEM-BEM approaches, the proposed method reduces BEM unknowns by approximately 30%, leading to runtime reductions of up to 46% in scattering cases, while maintaining comparable accuracy.