In this communication, we present an all-metal circularly polarized (CP) endfire antenna array composed of a series of perturbed stubs and shared tightly coupled resonant elements integrated with a double-sided parallel stripline (DSPSL). By feeding one port with an SMA connector while terminating the other with a $50~Omega $ load, the antenna can achieve left-hand/right-hand circular polarization (LHCP/RHCP) radiation in the endfire direction. When the excitation and the matched load exchange positions, it can also generate RHCP/LHCP radiation. A prototype was fabricated to validate the design, and measurements conducted in an anechoic chamber demonstrated excellent agreement with simulation results. The proposed antenna exhibits an impedance bandwidth and axial ratio (AR) bandwidth of 8%, an efficient of 82%91%, and an endfire gains of 12.2 dBi for LHCP and 11.9 dBi for RHCP. The compact size $(3.5lambda _{0} times 0.58lambda _{0})$ of the antenna further enhances its practicality and makes it a promising candidate for long-distance and fixed-point communication systems.
在本次通信中，我们提出了一种全金属圆极化（CP）端射天线阵列，该阵列由一系列摄动存根和共享紧密耦合谐振元件组成，并集成了双面平行带状线（DSPSL）。通过在一个端口上插入SMA连接器，同时在另一个端口上加载$50~Omega $负载，天线可以在端射方向上实现左/右圆极化（LHCP/RHCP）辐射。当励磁与匹配负载交换位置时，也能产生RHCP/LHCP辐射。制作了一个原型来验证设计，并在消声室中进行了测量，结果与仿真结果非常吻合。该天线的阻抗带宽和轴向比带宽均为8%, an efficient of 82%91%, and an endfire gains of 12.2 dBi for LHCP and 11.9 dBi for RHCP. The compact size $(3.5lambda _{0} times 0.58lambda _{0})$ of the antenna further enhances its practicality and makes it a promising candidate for long-distance and fixed-point communication systems.
{"title":"All-Metal High Gain and Circularly Polarized Endfire Antenna Based on Perturbed Horizontal Stubs and Shared Tightly Coupled Dipoles","authors":"Zhe Wu;Fantao Wu;Jian Xu;Yu Yun;Lingkun Ma;Nengwu Liu;Ying Liu;Tao Tao;Ling Sun;Ruiqi Wang;Yindi Wang","doi":"10.1109/TAP.2025.3632332","DOIUrl":"https://doi.org/10.1109/TAP.2025.3632332","url":null,"abstract":"In this communication, we present an all-metal circularly polarized (CP) endfire antenna array composed of a series of perturbed stubs and shared tightly coupled resonant elements integrated with a double-sided parallel stripline (DSPSL). By feeding one port with an SMA connector while terminating the other with a <inline-formula> <tex-math>$50~Omega $ </tex-math></inline-formula> load, the antenna can achieve left-hand/right-hand circular polarization (LHCP/RHCP) radiation in the endfire direction. When the excitation and the matched load exchange positions, it can also generate RHCP/LHCP radiation. A prototype was fabricated to validate the design, and measurements conducted in an anechoic chamber demonstrated excellent agreement with simulation results. The proposed antenna exhibits an impedance bandwidth and axial ratio (AR) bandwidth of 8%, an efficient of 82%91%, and an endfire gains of 12.2 dBi for LHCP and 11.9 dBi for RHCP. The compact size <inline-formula> <tex-math>$(3.5lambda _{0} times 0.58lambda _{0})$ </tex-math></inline-formula> of the antenna further enhances its practicality and makes it a promising candidate for long-distance and fixed-point communication systems.","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"74 2","pages":"2119-2124"},"PeriodicalIF":5.8,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146199237","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This work presents a gradual interval mapping method (IMM) integration framework for the rapid design of multiple metasurfaces. By leveraging the deep neural networks, the gradual design framework efficiently predicts both S-parameters and geometric structures using interval folding lines as input. Unlike traditional and current approaches, the proposed method integrates IMM with a filling strategy, simplifying data processing and enhancing prediction accuracy. To validate its effectiveness, three types of frequency-selective surface (FSS) are designed, one of which is fabricated and experimentally measured in a microwave chamber. Simulation and measurement results confirm the superior performance of the framework, highlighting its potential for intelligent electromagnetic device design. Finally, we conclude with a short discussion of the work, including its limitations and developability.
{"title":"The Gradual Interval Mapping Method Integration Framework for Multiple Metasurfaces Fast Design","authors":"Peng Wang;Chao Luo;Zhenning Li;Wen Jiang;Tao Hong;Gert Frølund Pedersen;Ming Shen","doi":"10.1109/TAP.2025.3632301","DOIUrl":"https://doi.org/10.1109/TAP.2025.3632301","url":null,"abstract":"This work presents a gradual interval mapping method (IMM) integration framework for the rapid design of multiple metasurfaces. By leveraging the deep neural networks, the gradual design framework efficiently predicts both S-parameters and geometric structures using interval folding lines as input. Unlike traditional and current approaches, the proposed method integrates IMM with a filling strategy, simplifying data processing and enhancing prediction accuracy. To validate its effectiveness, three types of frequency-selective surface (FSS) are designed, one of which is fabricated and experimentally measured in a microwave chamber. Simulation and measurement results confirm the superior performance of the framework, highlighting its potential for intelligent electromagnetic device design. Finally, we conclude with a short discussion of the work, including its limitations and developability.","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"74 2","pages":"2167-2172"},"PeriodicalIF":5.8,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146199041","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-14DOI: 10.1109/TAP.2025.3630887
Sicheng An;Luca Di Rienzo;Hao Qin;Xiaojie Zhu;Xingqi Zhang;Lorenzo Codecasa
Radio wave propagation modeling in tunnels is crucial to designing reliable wireless communication systems. Among the techniques available, the parabolic wave equation (PWE) methods have been widely utilized, due to their balance of accuracy and efficiency. However, the accuracy of the PWE methods depends on precise knowledge of tunnel environments, which are subject to uncertainties. While Monte Carlo (MC) methods are reliable for uncertainty analysis, they are computationally intensive. Polynomial chaos expansion (PCE) methods, though efficient, struggle with high-dimensional inputs. This communication applies the multilevel MC (MLMC) method to the PWE method in a nonintrusive way. MLMC is employed to address uncertainties arising from various sources. Such an MLMC-PWE method provides efficient estimations of the mean and variance of quantities of interest (QoI) by utilizing a multiscale hierarchy of spatial discretization. Numerical examples across different tunnel geometries demonstrate that the MLMC-PWE method achieves lower computational costs and improves efficiency relative to the MC-PWE method and the PCE-PWE method.
{"title":"Multilevel Monte Carlo Coupled With the Parabolic Wave Equation Method for Uncertainty Analysis of Radio Wave Propagation in Tunnels","authors":"Sicheng An;Luca Di Rienzo;Hao Qin;Xiaojie Zhu;Xingqi Zhang;Lorenzo Codecasa","doi":"10.1109/TAP.2025.3630887","DOIUrl":"https://doi.org/10.1109/TAP.2025.3630887","url":null,"abstract":"Radio wave propagation modeling in tunnels is crucial to designing reliable wireless communication systems. Among the techniques available, the parabolic wave equation (PWE) methods have been widely utilized, due to their balance of accuracy and efficiency. However, the accuracy of the PWE methods depends on precise knowledge of tunnel environments, which are subject to uncertainties. While Monte Carlo (MC) methods are reliable for uncertainty analysis, they are computationally intensive. Polynomial chaos expansion (PCE) methods, though efficient, struggle with high-dimensional inputs. This communication applies the multilevel MC (MLMC) method to the PWE method in a nonintrusive way. MLMC is employed to address uncertainties arising from various sources. Such an MLMC-PWE method provides efficient estimations of the mean and variance of quantities of interest (QoI) by utilizing a multiscale hierarchy of spatial discretization. Numerical examples across different tunnel geometries demonstrate that the MLMC-PWE method achieves lower computational costs and improves efficiency relative to the MC-PWE method and the PCE-PWE method.","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"74 2","pages":"2191-2196"},"PeriodicalIF":5.8,"publicationDate":"2025-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146199212","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-10DOI: 10.1109/TAP.2025.3628959
Yanhong Xu;Heming Xie;Can Cui;Jingwei Xu;Xin Dai
To address the issue of the iterative Fourier transform (IFT) method easily falling into local optima in thinned array synthesis, this communication proposes a two-stage alternating IFT (TSAIFT) approach for large thinned planar circular array synthesis, which involves an optimization stage utilizing a modified IFT density tapering (IFTDT), and a relaxation stage employing the IFT with constrained amplitude range. Specifically, the dynamic reference excitation mechanism is proposed to enhance the global exploration ability, which allows the number of active array elements in each ring to be iteratively adjusted during the optimization stage. In the relaxation stage, the excitation amplitudes are no longer constrained to be 0 or 1. Instead, a relaxation factor is introduced to gradually accumulate the small variations in the excitation coefficients. When the optimization stage gets trapped in local optima, it transitions to the relaxation stage to escape from the local optima. The two stages alternate to approach the optimal solution gradually. Due to the adoption of the low-computational IFT for both optimization and escaping from local optima, the proposed TSAIFT is capable of achieving competitive optimization results within a relatively short period of time, even when dealing with large-scale arrays. Several typical thinned planar circular arrays formed by isotropic elements and microstrip patch elements are synthesized, and the experimental results demonstrate the effectiveness of the proposed approach in optimization performance and computational efficiency.
{"title":"Synthesis of Large Thinned Planar Circular Array Using Two-Stage Alternating Iterative FFT","authors":"Yanhong Xu;Heming Xie;Can Cui;Jingwei Xu;Xin Dai","doi":"10.1109/TAP.2025.3628959","DOIUrl":"https://doi.org/10.1109/TAP.2025.3628959","url":null,"abstract":"To address the issue of the iterative Fourier transform (IFT) method easily falling into local optima in thinned array synthesis, this communication proposes a two-stage alternating IFT (TSAIFT) approach for large thinned planar circular array synthesis, which involves an optimization stage utilizing a modified IFT density tapering (IFTDT), and a relaxation stage employing the IFT with constrained amplitude range. Specifically, the dynamic reference excitation mechanism is proposed to enhance the global exploration ability, which allows the number of active array elements in each ring to be iteratively adjusted during the optimization stage. In the relaxation stage, the excitation amplitudes are no longer constrained to be 0 or 1. Instead, a relaxation factor is introduced to gradually accumulate the small variations in the excitation coefficients. When the optimization stage gets trapped in local optima, it transitions to the relaxation stage to escape from the local optima. The two stages alternate to approach the optimal solution gradually. Due to the adoption of the low-computational IFT for both optimization and escaping from local optima, the proposed TSAIFT is capable of achieving competitive optimization results within a relatively short period of time, even when dealing with large-scale arrays. Several typical thinned planar circular arrays formed by isotropic elements and microstrip patch elements are synthesized, and the experimental results demonstrate the effectiveness of the proposed approach in optimization performance and computational efficiency.","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"74 2","pages":"2155-2160"},"PeriodicalIF":5.8,"publicationDate":"2025-11-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146199247","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-11-07DOI: 10.1109/TAP.2025.3628054
Qi Zou;Pei Xiao;Yu Lin;Huanhuan Peng;Zhuolin Deng;Gui Gao;Gaosheng Li
An innovative ultralow-profile dual circularly polarization (DCP) beam-reconfigurable array antenna based on the Risley prism beam-scanning principle (RPSP), operating in the L-band, is proposed in this communication. An upgraded version of RPSP combines a large aperture feed array antenna and a phase-gradient transmissive metasurface (TM), both featuring linear gradient phase (LGP) properties, to achieve the ultralow profile characteristics. Additionally, a passive DCP TM operating in the L-band is designed for the first time, offering 360° phase control and high transmission efficiency. The large array feed antenna is implemented using an $8times 8$ array of stacked microstrip patches, achieving DCP radiation based on 3 dB couplers. By mechanically rotating the two array layers, the beam can achieve elevation angle scanning from 0° to 48° and azimuth angle scanning from 0° to 360° for DCP radiation. The measured maximum aperture efficiency (AE) reaches 37.63% at 1.65 GHz, with the antenna profile height being 77 mm ($0.41lambda _{0}$ ). The proposed DCP-RPA features an ultralow profile, low-cost manufacturability, and low loss, making it highly suitable for applications in weather radar, satellite communications, and next-generation mobile communication.
提出了一种基于Risley棱镜波束扫描原理（RPSP）的超低轮廓双圆极化波束可重构阵列天线，工作在l波段。升级版本的RPSP结合了大孔径馈电阵列天线和相位梯度传输超表面（TM），两者都具有线性梯度相位（LGP）特性，以实现超低轮廓特性。此外，还首次设计了工作在l波段的无源DCP TM，提供360°相位控制和高传输效率。大阵列馈电天线采用$8 × 8$的堆叠微带贴片阵列，实现基于3db耦合器的DCP辐射。通过机械旋转两层阵列，波束可以对DCP辐射进行0°~ 48°仰角扫描和0°~ 360°方位角扫描。在1.65 GHz时，测量到的最大孔径效率（AE）达到37.63%，天线轮廓高度为77 mm （$0.41lambda _{0}$）。提议的DCP-RPA具有超低外形、低制造成本和低损耗的特点，使其非常适合应用于气象雷达、卫星通信和下一代移动通信。
{"title":"An Ultralow-Profile Dual Circularly Polarized Beam-Scanning Array Antenna Based on Risley Prism at L-Band","authors":"Qi Zou;Pei Xiao;Yu Lin;Huanhuan Peng;Zhuolin Deng;Gui Gao;Gaosheng Li","doi":"10.1109/TAP.2025.3628054","DOIUrl":"https://doi.org/10.1109/TAP.2025.3628054","url":null,"abstract":"An innovative ultralow-profile dual circularly polarization (DCP) beam-reconfigurable array antenna based on the Risley prism beam-scanning principle (RPSP), operating in the L-band, is proposed in this communication. An upgraded version of RPSP combines a large aperture feed array antenna and a phase-gradient transmissive metasurface (TM), both featuring linear gradient phase (LGP) properties, to achieve the ultralow profile characteristics. Additionally, a passive DCP TM operating in the L-band is designed for the first time, offering 360° phase control and high transmission efficiency. The large array feed antenna is implemented using an <inline-formula> <tex-math>$8times 8$ </tex-math></inline-formula> array of stacked microstrip patches, achieving DCP radiation based on 3 dB couplers. By mechanically rotating the two array layers, the beam can achieve elevation angle scanning from 0° to 48° and azimuth angle scanning from 0° to 360° for DCP radiation. The measured maximum aperture efficiency (AE) reaches 37.63% at 1.65 GHz, with the antenna profile height being 77 mm (<inline-formula> <tex-math>$0.41lambda _{0}$ </tex-math></inline-formula>). The proposed DCP-RPA features an ultralow profile, low-cost manufacturability, and low loss, making it highly suitable for applications in weather radar, satellite communications, and next-generation mobile communication.","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"74 2","pages":"2161-2166"},"PeriodicalIF":5.8,"publicationDate":"2025-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146199044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-31DOI: 10.1109/TAP.2025.3622438
{"title":"IEEE Transactions on Antennas and Propagation Information for Authors","authors":"","doi":"10.1109/TAP.2025.3622438","DOIUrl":"https://doi.org/10.1109/TAP.2025.3622438","url":null,"abstract":"","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"73 11","pages":"C3-C3"},"PeriodicalIF":5.8,"publicationDate":"2025-10-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11223719","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145405323","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-31DOI: 10.1109/TAP.2025.3622440
{"title":"Institutional Listings","authors":"","doi":"10.1109/TAP.2025.3622440","DOIUrl":"https://doi.org/10.1109/TAP.2025.3622440","url":null,"abstract":"","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"73 11","pages":"C4-C4"},"PeriodicalIF":5.8,"publicationDate":"2025-10-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11223718","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145405321","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-31DOI: 10.1109/TAP.2025.3622436
{"title":"Recent Advances in Synthetic Aperture Antennas: Design, Modelling, and Measurement","authors":"","doi":"10.1109/TAP.2025.3622436","DOIUrl":"https://doi.org/10.1109/TAP.2025.3622436","url":null,"abstract":"","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"73 11","pages":"9672-9673"},"PeriodicalIF":5.8,"publicationDate":"2025-10-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11223715","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145405415","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-10-30DOI: 10.1109/TAP.2025.3625189
Guangnan Xing;Shiyong Li;Ahmad Hoorfar;Qiang An;Guoqiang Zhao;Houjun Sun
In millimeter-wave (MMW) imaging, the focusing performance of sparse multiple-input-multiple-output (MIMO) arrays is susceptible to various noise interferences during the propagation of electromagnetic waves. In addition, the sparse array imaging based on the traditional compressive sensing (CS) algorithms suffers from low computational efficiency due to the numerous iterations, especially in 3-D imaging scenarios. To address these limitations, an interpretable self-supervised network with a parallel structure is proposed for fast enhanced imaging via sparse cross MIMO arrays. The designed interpolation-free MIMO operators are embedded into a generalized complex-valued framework unfolded by the alternating direction method of multipliers (ADMMs) to form the encoder of reconstruction branch, avoiding large-scale MIMO sensing matrix operations and allowing for the fast 3-D imaging processing of sparse MIMO data. The decoder is constructed to map the data from image domain to signal domain, thereby producing the pseudo-labels for self-supervised learning and eliminating the need for costly data annotation. The enhancement branch employs an optimization module based on the complex-valued total variation (CTV) and $l_{1}$ norms to guide the network training, thus improving the noise immunity and imaging performance of the network. The simulations and experiments show that the proposed network can obtain better-focused 3-D imaging results in a short time.
{"title":"Self-Supervised Enhanced Imaging Network for Sparse Cross MIMO Arrays","authors":"Guangnan Xing;Shiyong Li;Ahmad Hoorfar;Qiang An;Guoqiang Zhao;Houjun Sun","doi":"10.1109/TAP.2025.3625189","DOIUrl":"https://doi.org/10.1109/TAP.2025.3625189","url":null,"abstract":"In millimeter-wave (MMW) imaging, the focusing performance of sparse multiple-input-multiple-output (MIMO) arrays is susceptible to various noise interferences during the propagation of electromagnetic waves. In addition, the sparse array imaging based on the traditional compressive sensing (CS) algorithms suffers from low computational efficiency due to the numerous iterations, especially in 3-D imaging scenarios. To address these limitations, an interpretable self-supervised network with a parallel structure is proposed for fast enhanced imaging via sparse cross MIMO arrays. The designed interpolation-free MIMO operators are embedded into a generalized complex-valued framework unfolded by the alternating direction method of multipliers (ADMMs) to form the encoder of reconstruction branch, avoiding large-scale MIMO sensing matrix operations and allowing for the fast 3-D imaging processing of sparse MIMO data. The decoder is constructed to map the data from image domain to signal domain, thereby producing the pseudo-labels for self-supervised learning and eliminating the need for costly data annotation. The enhancement branch employs an optimization module based on the complex-valued total variation (CTV) and <inline-formula> <tex-math>$l_{1}$ </tex-math></inline-formula> norms to guide the network training, thus improving the noise immunity and imaging performance of the network. The simulations and experiments show that the proposed network can obtain better-focused 3-D imaging results in a short time.","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"74 1","pages":"1281-1286"},"PeriodicalIF":5.8,"publicationDate":"2025-10-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146015924","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Artificial intelligence (AI) for computational electromagnetics (CEM) has drawn an increasing attention in recent years. However, current AI-based CEM methods face two key challenges: the high production cost of training datasets and poor generalization ability of trained neural networks (NNs). In this communication, we employ active learning (AL) and incremental learning (IL) to tackle the issues. AL cuts the dataset production cost by selecting the most informative samples before NN training, guided by evaluating how well the predicted results by a pretrained NN satisfy the boundary conditions or governing equations. By this way, the dataset generation time and storage are significantly reduced compared to fully supervised learning (SL). IL improves the generalization ability by enabling an NN pretrained for one type of targets to efficiently acquire knowledge for a new type of targets while retaining prior learning. This strategy is substantially more efficient than training new NNs from scratch for each type of targets. The effectiveness of the proposed AL-IL framework is demonstrated for 3-D conducting, dielectric, and metal–dielectric composite targets. It establishes a valuable paradigm for AI-based CEM, accelerating dataset generation and enhancing model generalization.
{"title":"AI for Evaluation of Electromagnetic Scattering Using Active and Incremental Learning","authors":"De-Hua Kong;Wen-Wei Zhang;Jia-Qi Kang;Wen-Chi Huang;Jia-Ning Cao;Xing-Yue Guo;Ming-Yao Xia","doi":"10.1109/TAP.2025.3624223","DOIUrl":"https://doi.org/10.1109/TAP.2025.3624223","url":null,"abstract":"Artificial intelligence (AI) for computational electromagnetics (CEM) has drawn an increasing attention in recent years. However, current AI-based CEM methods face two key challenges: the high production cost of training datasets and poor generalization ability of trained neural networks (NNs). In this communication, we employ active learning (AL) and incremental learning (IL) to tackle the issues. AL cuts the dataset production cost by selecting the most informative samples before NN training, guided by evaluating how well the predicted results by a pretrained NN satisfy the boundary conditions or governing equations. By this way, the dataset generation time and storage are significantly reduced compared to fully supervised learning (SL). IL improves the generalization ability by enabling an NN pretrained for one type of targets to efficiently acquire knowledge for a new type of targets while retaining prior learning. This strategy is substantially more efficient than training new NNs from scratch for each type of targets. The effectiveness of the proposed AL-IL framework is demonstrated for 3-D conducting, dielectric, and metal–dielectric composite targets. It establishes a valuable paradigm for AI-based CEM, accelerating dataset generation and enhancing model generalization.","PeriodicalId":13102,"journal":{"name":"IEEE Transactions on Antennas and Propagation","volume":"74 1","pages":"1245-1250"},"PeriodicalIF":5.8,"publicationDate":"2025-10-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146015926","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：