Pub Date : 2025-01-13DOI: 10.1109/JMMCT.2025.3528484
Xiaofan Jia;Mingyu Wang;Qiqi Dai;Chao-Fu Wang;Abdulkadir C. Yucel
A physics-informed deep learning-based scheme is introduced for computing partial inductances of interconnects. This scheme takes a physics-based skin depth map and a geometry identifier of the interconnects as inputs and provides the current density distribution on the interconnects as the output. The predicted currents are then used to compute the partial self-resistances, self-inductances, and mutual-inductances of the interconnects. The proposed method leverages an Attention U-net, a U-shaped convolutional neural network with attention modules. During the training of Attention U-net, a specifically designed loss function is used to ensure the accurate modeling of the currents on the structure as well as ports. The accuracy, efficiency, and generalization ability of this physics-informed deep learning method are demonstrated via inductance extraction of the interconnects with and without a ground plane, including straight single interconnects, interconnects with sharp bends, parallel interconnects, and multiple conductor crossover buses. Numerical results show that the proposed scheme can predict the current density distribution of one interconnect scenario in 15.63 ms on GPU, 1157x faster than the physics-based solver, while providing self-inductances, mutual-inductances, and self-resistances of interconnects with around 1%, 3%, and 4% ${{ell }_2}$-norm error, respectively.
{"title":"Deep Learning-Based Partial Inductance Extraction of 3-D Interconnects","authors":"Xiaofan Jia;Mingyu Wang;Qiqi Dai;Chao-Fu Wang;Abdulkadir C. Yucel","doi":"10.1109/JMMCT.2025.3528484","DOIUrl":"https://doi.org/10.1109/JMMCT.2025.3528484","url":null,"abstract":"A physics-informed deep learning-based scheme is introduced for computing partial inductances of interconnects. This scheme takes a physics-based skin depth map and a geometry identifier of the interconnects as inputs and provides the current density distribution on the interconnects as the output. The predicted currents are then used to compute the partial self-resistances, self-inductances, and mutual-inductances of the interconnects. The proposed method leverages an Attention U-net, a U-shaped convolutional neural network with attention modules. During the training of Attention U-net, a specifically designed loss function is used to ensure the accurate modeling of the currents on the structure as well as ports. The accuracy, efficiency, and generalization ability of this physics-informed deep learning method are demonstrated via inductance extraction of the interconnects with and without a ground plane, including straight single interconnects, interconnects with sharp bends, parallel interconnects, and multiple conductor crossover buses. Numerical results show that the proposed scheme can predict the current density distribution of one interconnect scenario in 15.63 ms on GPU, 1157x faster than the physics-based solver, while providing self-inductances, mutual-inductances, and self-resistances of interconnects with around 1%, 3%, and 4% <inline-formula><tex-math>${{ell }_2}$</tex-math></inline-formula>-norm error, respectively.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"10 ","pages":"112-124"},"PeriodicalIF":1.8,"publicationDate":"2025-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143360947","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-01-10DOI: 10.1109/JMMCT.2025.3528076
Sairam SD;Sriram Kumar Dhamodharan
A multi-layer perceptron (MLP) model was applied to electromagnetic shielding to analyze a coupled ring anisotropic frequency selective surface (CRAFSS) using an equivalent circuit model. The shielding structure, based on a single-sided RT 5880 array, features unit elements with dimensions of $0.55lambda _{0} times 0.41lambda _{0}$ at the resonant frequency. Various deep neural network (DNN) configurations with hidden layers were tested to achieve optimal results, reaching a minimal mean square error (MSE) of $1.012 times 10^{-4}$. The MLP was trained using input parameters such as S-parameters, resonant frequency, and shielding effectiveness, with the output being the dimensions of the proposed shielding structure. The dataset, built from capacitance and inductance values, was used for testing, training, and validation within the neural network, eventually employing inverse modeling for output prediction. The structure demonstrated stable bandwidth performance despite changes in the incidence angle of transverse magnetic (TM) and transverse electric (TE) polarizations, shifting from $theta$ = $0^{0}$ to $60^{0}$. The anisotropic FSS was developed and evaluated, with deep learning results and electromagnetic (EM) simulations playing a key role in the design process.
将多层感知器（MLP）模型应用于电磁屏蔽中，利用等效电路模型分析耦合环形各向异性频率选择面（CRAFSS）。该屏蔽结构基于单面RT 5880阵列，在谐振频率处具有尺寸为$0.55lambda _{0} times 0.41lambda _{0}$的单元元件。为了获得最优的结果，我们测试了各种带有隐藏层的深度神经网络（DNN）配置，其均方误差（MSE）最小值为$1.012 times 10^{-4}$。MLP使用s参数、谐振频率和屏蔽效率等输入参数进行训练，输出是所提出的屏蔽结构的尺寸。该数据集由电容和电感值构建，用于神经网络内的测试、训练和验证，最终采用逆建模进行输出预测。尽管横向磁极化（TM）和横向电极化（TE）的入射角从$theta$ = $0^{0}$变化到$60^{0}$，但该结构的带宽性能仍然稳定。开发和评估了各向异性FSS，深度学习结果和电磁（EM）模拟在设计过程中发挥了关键作用。
{"title":"EMI Shielding With Anisotropic Frequency Selective Surfaces: A Neural Network and Equivalent Circuit Approach","authors":"Sairam SD;Sriram Kumar Dhamodharan","doi":"10.1109/JMMCT.2025.3528076","DOIUrl":"https://doi.org/10.1109/JMMCT.2025.3528076","url":null,"abstract":"A multi-layer perceptron (MLP) model was applied to electromagnetic shielding to analyze a coupled ring anisotropic frequency selective surface (CRAFSS) using an equivalent circuit model. The shielding structure, based on a single-sided RT 5880 array, features unit elements with dimensions of <inline-formula><tex-math>$0.55lambda _{0} times 0.41lambda _{0}$</tex-math></inline-formula> at the resonant frequency. Various deep neural network (DNN) configurations with hidden layers were tested to achieve optimal results, reaching a minimal mean square error (MSE) of <inline-formula><tex-math>$1.012 times 10^{-4}$</tex-math></inline-formula>. The MLP was trained using input parameters such as S-parameters, resonant frequency, and shielding effectiveness, with the output being the dimensions of the proposed shielding structure. The dataset, built from capacitance and inductance values, was used for testing, training, and validation within the neural network, eventually employing inverse modeling for output prediction. The structure demonstrated stable bandwidth performance despite changes in the incidence angle of transverse magnetic (TM) and transverse electric (TE) polarizations, shifting from <inline-formula><tex-math>$theta$</tex-math></inline-formula> = <inline-formula><tex-math>$0^{0}$</tex-math></inline-formula> to <inline-formula><tex-math>$60^{0}$</tex-math></inline-formula>. The anisotropic FSS was developed and evaluated, with deep learning results and electromagnetic (EM) simulations playing a key role in the design process.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"10 ","pages":"94-103"},"PeriodicalIF":1.8,"publicationDate":"2025-01-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143105960","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-31DOI: 10.1109/JMMCT.2024.3524598
Yuxian Zhang;Yilin Kang;Naixing Feng;Xiaoli Feng;Zhixiang Huang;Atef Z. Elsherbeni
In this article, to breakthrough the constraint from conventional finite-difference time-domain (FDTD) method, we firstly propose a scale-compressed technique (SCT) working for the FDTD method, been called SCT-FDTD for short, to reduce three-dimensional (3-D) into one-dimensional (1-D) processes and capture the propagation coefficients. Combining with Maxwell's curl equations, the transverse wave vectors (kx, ky) can be defined as the fixed values, which let the curl operator become the curl matrix with only z-directional derivative. The obvious advantage demonstrated by above is that it does not require excessive computational processes to obtain high-dimensional numerical results with reasonable accuracy. By comparing with commercial software COMSOL by the TE/TM illumination in multi-layered biaxial anisotropy, those results from SCT-FDTD method are entirely consistent. More importantly, the SCT-FDTD possesses less CPU time and lower computational resources for COMSOL.
{"title":"Scale-Compressed Technique in Finite-Difference Time-Domain Method for Multi-Layered Anisotropic Media","authors":"Yuxian Zhang;Yilin Kang;Naixing Feng;Xiaoli Feng;Zhixiang Huang;Atef Z. Elsherbeni","doi":"10.1109/JMMCT.2024.3524598","DOIUrl":"https://doi.org/10.1109/JMMCT.2024.3524598","url":null,"abstract":"In this article, to breakthrough the constraint from conventional finite-difference time-domain (FDTD) method, we firstly propose a scale-compressed technique (SCT) working for the FDTD method, been called SCT-FDTD for short, to reduce three-dimensional (3-D) into one-dimensional (1-D) processes and capture the propagation coefficients. Combining with Maxwell's curl equations, the transverse wave vectors (<italic>k<sub>x</sub></i>, <italic>k<sub>y</sub></i>) can be defined as the fixed values, which let the curl operator become the curl matrix with only <italic>z</i>-directional derivative. The obvious advantage demonstrated by above is that it does not require excessive computational processes to obtain high-dimensional numerical results with reasonable accuracy. By comparing with commercial software COMSOL by the TE/TM illumination in multi-layered biaxial anisotropy, those results from SCT-FDTD method are entirely consistent. More importantly, the SCT-FDTD possesses less CPU time and lower computational resources for COMSOL.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"10 ","pages":"85-93"},"PeriodicalIF":1.8,"publicationDate":"2024-12-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142975849","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-12-20DOI: 10.1109/JMMCT.2024.3520488
Angelika S. Thalmayer;Keyu Xiao;Paul Wolff;Georg Fischer
The development of trustworthy simulation models is crucial for planning drug administration in magnetic drug targeting (MDT) interventions for future cancer treatment. In the MDT cancer therapy, the drug is bound to magnetic nanoparticles, which act as carriers and are guided through the cardiovascular system into the tumor region using an external magnetic field. Thus, the modeling represents a multiphysical problem and can be approached either by particle-based or concentration-based models. In this paper, both simulation approaches are implemented in COMSOL Multiphysics in a typical magnetic drug targeting scenario, verified by measurements, and compared among each other. Two different particle concentrations with and without an applied magnetic field of a Halbach array consisting of five permanent magnets in a tube flow system with a laminar velocity flow were investigated. Within this scope, an analytical model for calculating the system response for the detection of nanoparticles with a commercial susceptometer is derived, too. Considering the two implemented models and the investigated scenario, the concentration-based model shows a considerably better agreement with the experimental results for both with and without an applied magnetic field. The spatial resolution of the particle-based model is reduced due to the limited number of considered particles resulting in an inaccurate system response. Overall, the high number of new publications shows the need for research in this interdisciplinary research field to improve therapeutic success.
{"title":"Experimental and Numerical Modeling of Magnetic Drug Targeting: Can We Trust Particle-Based Models?","authors":"Angelika S. Thalmayer;Keyu Xiao;Paul Wolff;Georg Fischer","doi":"10.1109/JMMCT.2024.3520488","DOIUrl":"https://doi.org/10.1109/JMMCT.2024.3520488","url":null,"abstract":"The development of trustworthy simulation models is crucial for planning drug administration in magnetic drug targeting (MDT) interventions for future cancer treatment. In the MDT cancer therapy, the drug is bound to magnetic nanoparticles, which act as carriers and are guided through the cardiovascular system into the tumor region using an external magnetic field. Thus, the modeling represents a multiphysical problem and can be approached either by particle-based or concentration-based models. In this paper, both simulation approaches are implemented in COMSOL Multiphysics in a typical magnetic drug targeting scenario, verified by measurements, and compared among each other. Two different particle concentrations with and without an applied magnetic field of a Halbach array consisting of five permanent magnets in a tube flow system with a laminar velocity flow were investigated. Within this scope, an analytical model for calculating the system response for the detection of nanoparticles with a commercial susceptometer is derived, too. Considering the two implemented models and the investigated scenario, the concentration-based model shows a considerably better agreement with the experimental results for both with and without an applied magnetic field. The spatial resolution of the particle-based model is reduced due to the limited number of considered particles resulting in an inaccurate system response. Overall, the high number of new publications shows the need for research in this interdisciplinary research field to improve therapeutic success.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"10 ","pages":"69-84"},"PeriodicalIF":1.8,"publicationDate":"2024-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142938138","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Recently, there has been a surge in the development of data-driven propagation models. These models aspire to distill knowledge from propagation solvers or measured data and eventually become capable of predicting characteristics related to radiowave propagation. In this paper, we present the functionality of a generalizable and robust data-driven propagation model that enables efficient and reliable simulations of indoor wireless communication systems (IWCSs). In particular, we modify our previously introduced model, EM DeepRay, to consider the impact of antenna directivity, and we present a training and inference strategy that allows the simulation of large-scale and complicated IWCSs. Our data-driven model is trained over a rich data set comprising diverse building geometries, frequency bands, and antenna radiation patterns. Benchmarking its performance with that of a ray-tracer in complicated IWCSs with real-world measured data yields similar results that have a distinct advantage in terms of computational time. Ultimately, our work paves the way for replacing legacy IWCSs simulators, with high-fidelity artificial intelligence-based models.
{"title":"Rigorous Indoor Wireless Communication System Simulations With Deep Learning-Based Radio Propagation Models","authors":"Stefanos Bakirtzis;Kehai Qiu;Jiming Chen;Hui Song;Jie Zhang;Ian Wassell","doi":"10.1109/JMMCT.2024.3506693","DOIUrl":"https://doi.org/10.1109/JMMCT.2024.3506693","url":null,"abstract":"Recently, there has been a surge in the development of data-driven propagation models. These models aspire to distill knowledge from propagation solvers or measured data and eventually become capable of predicting characteristics related to radiowave propagation. In this paper, we present the functionality of a generalizable and robust data-driven propagation model that enables efficient and reliable simulations of indoor wireless communication systems (IWCSs). In particular, we modify our previously introduced model, EM DeepRay, to consider the impact of antenna directivity, and we present a training and inference strategy that allows the simulation of large-scale and complicated IWCSs. Our data-driven model is trained over a rich data set comprising diverse building geometries, frequency bands, and antenna radiation patterns. Benchmarking its performance with that of a ray-tracer in complicated IWCSs with real-world measured data yields similar results that have a distinct advantage in terms of computational time. Ultimately, our work paves the way for replacing legacy IWCSs simulators, with high-fidelity artificial intelligence-based models.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"10 ","pages":"58-68"},"PeriodicalIF":1.8,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142905782","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-29DOI: 10.1109/JMMCT.2024.3509773
Aggraj Gupta;Uday K Khankhoje
Deep learning frameworks are gaining prominence in the electromagnetics community for designing microwave and mm-wave devices. This paper presents a computationally efficient transfer learning technique for designing and scaling multi-band microstrip antennas to a desired dielectric and frequency of interest. The proposed methodology involves a two-step process. First, a pre-trained model trained extensively on air-filled microstrip antennas is used for knowledge transfer. This pre-trained model is fine-tuned with a limited set of dielectric simulations, reducing data acquisition costs. In the second step, the developed forward model serves as a surrogate to design dielectric-filled antennas using the Improved Binary Particle Swarm Optimization algorithm. In contrast to conventional methods, this approach enables the design of compact antennas across various dielectrics and frequency ranges, with a significantly reduced number of time-consuming dielectric simulations (88% fewer simulations) and a lower neural network training time (75% lesser time). We analyze the optimal ways of generating dielectric antenna datasets via scaling, and perform sensitivity analysis with respect to the antenna's physical parameters. We report simulation and experimental results for single and double band antennas fabricated using the proposed approach.
{"title":"Transfer Learning Based Rapid Design of Frequency and Dielectric Agile Antennas","authors":"Aggraj Gupta;Uday K Khankhoje","doi":"10.1109/JMMCT.2024.3509773","DOIUrl":"https://doi.org/10.1109/JMMCT.2024.3509773","url":null,"abstract":"Deep learning frameworks are gaining prominence in the electromagnetics community for designing microwave and mm-wave devices. This paper presents a computationally efficient transfer learning technique for designing and scaling multi-band microstrip antennas to a desired dielectric and frequency of interest. The proposed methodology involves a two-step process. First, a pre-trained model trained extensively on air-filled microstrip antennas is used for knowledge transfer. This pre-trained model is fine-tuned with a limited set of dielectric simulations, reducing data acquisition costs. In the second step, the developed forward model serves as a surrogate to design dielectric-filled antennas using the Improved Binary Particle Swarm Optimization algorithm. In contrast to conventional methods, this approach enables the design of compact antennas across various dielectrics and frequency ranges, with a significantly reduced number of time-consuming dielectric simulations (88% fewer simulations) and a lower neural network training time (75% lesser time). We analyze the optimal ways of generating dielectric antenna datasets via scaling, and perform sensitivity analysis with respect to the antenna's physical parameters. We report simulation and experimental results for single and double band antennas fabricated using the proposed approach.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"10 ","pages":"47-57"},"PeriodicalIF":1.8,"publicationDate":"2024-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142810340","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-20DOI: 10.1109/JMMCT.2024.3502830
Eng Leong Tan;Ding Yu Heh
This paper presents the critical-point-based stability analyses of finite-difference time-domain (FDTD) methods for Schrödinger equation incorporating vector and scalar potentials. Most previous FDTD formulations and stability analyses for the Schrödinger equation involve only the scalar potentials. On the other hand, the existing stability conditions that include both vector and scalar potentials were not thoroughly nor rigorously analyzed, hence they are inadequate for general cases. In this paper, rigorous stability analyses of the FDTD methods will be performed for Schrödinger equation in full 3D incorporating both vector and scalar potentials. New stability conditions are derived rigorously based on the critical points within the interior and boundary regions, while considering the local and global extrema across all variables. Two FDTD schemes are considered, of which one is updated entirely in complex form, and the other is decomposed into real and imaginary parts and updated in a leapfrog manner. Comparisons of the new stability conditions are made against those of prior works, highlighting the thoroughness, completeness and adequacy. Numerical experiments further validate the derived stability conditions and demonstrate their applicability in FDTD methods. Using these stability conditions, the FDTD methods are useful for simulations of quantum-electromagnetic interactions involving vector and scalar potentials.
{"title":"Critical-Point-Based Stability Analyses of Finite-Difference Time-Domain Methods for Schrödinger Equation Incorporating Vector and Scalar Potentials","authors":"Eng Leong Tan;Ding Yu Heh","doi":"10.1109/JMMCT.2024.3502830","DOIUrl":"https://doi.org/10.1109/JMMCT.2024.3502830","url":null,"abstract":"This paper presents the critical-point-based stability analyses of finite-difference time-domain (FDTD) methods for Schrödinger equation incorporating vector and scalar potentials. Most previous FDTD formulations and stability analyses for the Schrödinger equation involve only the scalar potentials. On the other hand, the existing stability conditions that include both vector and scalar potentials were not thoroughly nor rigorously analyzed, hence they are inadequate for general cases. In this paper, rigorous stability analyses of the FDTD methods will be performed for Schrödinger equation in full 3D incorporating both vector and scalar potentials. New stability conditions are derived rigorously based on the critical points within the interior and boundary regions, while considering the local and global extrema across all variables. Two FDTD schemes are considered, of which one is updated entirely in complex form, and the other is decomposed into real and imaginary parts and updated in a leapfrog manner. Comparisons of the new stability conditions are made against those of prior works, highlighting the thoroughness, completeness and adequacy. Numerical experiments further validate the derived stability conditions and demonstrate their applicability in FDTD methods. Using these stability conditions, the FDTD methods are useful for simulations of quantum-electromagnetic interactions involving vector and scalar potentials.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"10 ","pages":"38-46"},"PeriodicalIF":1.8,"publicationDate":"2024-11-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142798015","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-19DOI: 10.1109/JMMCT.2024.3502062
Yanan Liu;Hongliang Li;Jian-Ming Jin
In this paper, we present a machine learning technique based on analytic extension of eigenvalues and neural networks for the efficient modeling of high-frequency devices. In the proposed method, neural networks are used to learn the mapping between device's geometry and its modal equivalent circuit parameters. These circuit parameters are extracted from the eigen-decomposition of the deviceâs $Z$ -parameters at a few sample frequencies. The eigenvalues and eigenvectors of the $Z$ -matrix are analytically extended to other frequencies based on functional equations constructed from the lumped equivalent circuit model, from which the full electromagnetic response can be recovered. In addition to fully-connected neural network layers, our proposed model introduces an analytical projection branch based on AEE principles to maximize the information gain from samples in the training dataset. To improve the robustness and efficiency of the learning process, we introduce an adaptive gradient update algorithm. The overall model is end-to-end differentiable and can be integrated into gradient-based optimization methods. Numerical examples are provided to demonstrate the capability of the proposed method. Compared with traditional neural network-based models, the proposed approach achieves higher accuracy using fewer data samples and generalizes better to out-of-domain inputs.
{"title":"Physics-Informed Machine Learning for the Efficient Modeling of High-Frequency Devices","authors":"Yanan Liu;Hongliang Li;Jian-Ming Jin","doi":"10.1109/JMMCT.2024.3502062","DOIUrl":"https://doi.org/10.1109/JMMCT.2024.3502062","url":null,"abstract":"In this paper, we present a machine learning technique based on analytic extension of eigenvalues and neural networks for the efficient modeling of high-frequency devices. In the proposed method, neural networks are used to learn the mapping between device's geometry and its modal equivalent circuit parameters. These circuit parameters are extracted from the eigen-decomposition of the deviceâs \u0000<inline-formula><tex-math>$Z$</tex-math></inline-formula>\u0000-parameters at a few sample frequencies. The eigenvalues and eigenvectors of the \u0000<inline-formula><tex-math>$Z$</tex-math></inline-formula>\u0000-matrix are analytically extended to other frequencies based on functional equations constructed from the lumped equivalent circuit model, from which the full electromagnetic response can be recovered. In addition to fully-connected neural network layers, our proposed model introduces an analytical projection branch based on AEE principles to maximize the information gain from samples in the training dataset. To improve the robustness and efficiency of the learning process, we introduce an adaptive gradient update algorithm. The overall model is end-to-end differentiable and can be integrated into gradient-based optimization methods. Numerical examples are provided to demonstrate the capability of the proposed method. Compared with traditional neural network-based models, the proposed approach achieves higher accuracy using fewer data samples and generalizes better to out-of-domain inputs.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"10 ","pages":"28-37"},"PeriodicalIF":1.8,"publicationDate":"2024-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142798028","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2024-11-15DOI: 10.1109/JMMCT.2024.3499369
Jesse Rivera;John Blaske;Zhi Yao;Ruoda Zheng;Gregory P. Carman;Yuanxun Ethan Wang
Antenna miniaturization is currently facing increased performance demands while simultaneously lacking a computational framework to drive robust designs. Future platforms must radiate farther, at lower frequency, and be increasingly compact. While mechanical resonance based piezoelectric antenna arrays are a viable candidate, detrimental mutual depolarization effects arise that must be characterized by multi-scale simulations, coupling the elastodynamic and EM wave physics. This work presents an algorithm capable of performing such full-wave simulations to provide design guidance to engineers wishing to mitigate mutual depolarization. The relevant dynamic systems of equations are discretized and put into a Finite Difference Time Domain (FDTD) scheme. This scheme exhibits electrodynamic unconditional stability and features heavily graded meshes to directly tackle the time and length scale disparity between the mechanical and EM waves. The algorithm was validated by comparison with experimental data and analytical solutions. Additionally, the algorithm compared well with predicted values for depolarization. Simulations demonstrated that spacing within piezoelectric antenna arrays should not be made too small, as to induce undue mutual depolarization, or too large, as to not allow sufficient elements to contribute to array dipole moment. Computational guidance is also provided based on the authors’ own experiences.
{"title":"Mechanical Antenna Simulations via FDTD to Characterize Mutual Depolarization","authors":"Jesse Rivera;John Blaske;Zhi Yao;Ruoda Zheng;Gregory P. Carman;Yuanxun Ethan Wang","doi":"10.1109/JMMCT.2024.3499369","DOIUrl":"https://doi.org/10.1109/JMMCT.2024.3499369","url":null,"abstract":"Antenna miniaturization is currently facing increased performance demands while simultaneously lacking a computational framework to drive robust designs. Future platforms must radiate farther, at lower frequency, and be increasingly compact. While mechanical resonance based piezoelectric antenna arrays are a viable candidate, detrimental mutual depolarization effects arise that must be characterized by multi-scale simulations, coupling the elastodynamic and EM wave physics. This work presents an algorithm capable of performing such full-wave simulations to provide design guidance to engineers wishing to mitigate mutual depolarization. The relevant dynamic systems of equations are discretized and put into a Finite Difference Time Domain (FDTD) scheme. This scheme exhibits electrodynamic unconditional stability and features heavily graded meshes to directly tackle the time and length scale disparity between the mechanical and EM waves. The algorithm was validated by comparison with experimental data and analytical solutions. Additionally, the algorithm compared well with predicted values for depolarization. Simulations demonstrated that spacing within piezoelectric antenna arrays should not be made too small, as to induce undue mutual depolarization, or too large, as to not allow sufficient elements to contribute to array dipole moment. Computational guidance is also provided based on the authors’ own experiences.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"10 ","pages":"8-27"},"PeriodicalIF":1.8,"publicationDate":"2024-11-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142777740","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Wireless communication technology has become important in modern life. Real-world radio environments present significant challenges, particularly concerning latency and multipath fading. A promising solution is represented by reconfigurable intelligent surfaces (RIS), which can manipulate electromagnetic waves to enhance transmission quality. In this study, we introduce a novel approach that employs the quantum approximate optimization algorithm (QAOA) to efficiently configure RIS in multipath environments. Applying the spin glass (SG) theoretical framework to describe chaotic systems, along with a variable noise model, we propose a quantum-based minimization algorithm to optimize RIS in various electromagnetic scenarios affected by multipath fading. The method involves training a parameterized quantum circuit using a mathematical model that scales with the size of the RIS. When applied to different EM scenarios, it directly identifies the optimal RIS configuration. This approach eliminates the need for large datasets for training, validation, and testing, streamlines, and accelerates the training process. Furthermore, the algorithm will not need to be rerun for each individual scenario. In particular, our analysis considers a system with one transmitting antenna, multiple receiving antennas, and varying noise levels. The results show that QAOA enhances the performance of RIS in both noise-free and noisy environments, highlighting the potential of quantum computing to address the complexities of RIS optimization and improve the performance of the wireless network.
{"title":"Quantum Optimization of Reconfigurable Intelligent Surfaces for Mitigating Multipath Fading in Wireless Networks","authors":"Emanuel Colella;Luca Bastianelli;Valter Mariani Primiani;Zhen Peng;Franco Moglie;Gabriele Gradoni","doi":"10.1109/JMMCT.2024.3494037","DOIUrl":"https://doi.org/10.1109/JMMCT.2024.3494037","url":null,"abstract":"Wireless communication technology has become important in modern life. Real-world radio environments present significant challenges, particularly concerning latency and multipath fading. A promising solution is represented by reconfigurable intelligent surfaces (RIS), which can manipulate electromagnetic waves to enhance transmission quality. In this study, we introduce a novel approach that employs the quantum approximate optimization algorithm (QAOA) to efficiently configure RIS in multipath environments. Applying the spin glass (SG) theoretical framework to describe chaotic systems, along with a variable noise model, we propose a quantum-based minimization algorithm to optimize RIS in various electromagnetic scenarios affected by multipath fading. The method involves training a parameterized quantum circuit using a mathematical model that scales with the size of the RIS. When applied to different EM scenarios, it directly identifies the optimal RIS configuration. This approach eliminates the need for large datasets for training, validation, and testing, streamlines, and accelerates the training process. Furthermore, the algorithm will not need to be rerun for each individual scenario. In particular, our analysis considers a system with one transmitting antenna, multiple receiving antennas, and varying noise levels. The results show that QAOA enhances the performance of RIS in both noise-free and noisy environments, highlighting the potential of quantum computing to address the complexities of RIS optimization and improve the performance of the wireless network.","PeriodicalId":52176,"journal":{"name":"IEEE Journal on Multiscale and Multiphysics Computational Techniques","volume":"9 ","pages":"403-414"},"PeriodicalIF":1.8,"publicationDate":"2024-11-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=10747251","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142736643","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","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：