The proposed approach introduces an innovative feeding mechanism enabling efficient directive radiation with a planar antenna array. A microstrip T-type power divider is employed for initial power distribution, while two symmetrically placed slots are utilized to achieve direct mode conversion into the higher-order TE 50 mode within the SIW structure. A prototype was fabricated to validate the concept experimentally. The measured results demonstrate excellent agreement with simulations, confirming a fractional impedance bandwidth of 3.4% (5.70–5.90 GHz for S 11 < − 10 dB) and a peak realized gain of 10.2 dBi at 5.8 GHz, with a measured radiation efficiency of 87.2%. This design offers significant advantages, including simplified fabrication, reduced production costs, and enhanced tolerance to manufacturing variations, making it well-suited for communication systems at 5.8 GHz requiring directive radiation.
The design and optimization of large antenna arrays and Reconfigurable Intelligent Surfaces (RIS) have become critical for emerging wireless communication systems, including 5G and 6G networks. Although technologies such as Massive Multiple Input Multiple Output (MIMO) and RIS offer unprecedented opportunities for enhanced coverage, improved spectral efficiency, and adaptive propagation environments, the corresponding electromagnetic structures often present extremely large and complex design spaces that require exhaustive exploration to achieve optimal performance. As a result, achieving goals such as miniaturization, bandwidth enhancement, improved scattering characteristics, or precise phase control becomes a major focus when working with these complex spaces. In this paper, we introduce a systematic methodology for developing an optimization framework dedicated to planar electromagnetic structures using binary pixelization of the conducting surface. By discretizing the design region into a matrix of metallic and non-metallic pixels, the approach enables flexible topology manipulation while maintaining compatibility with full-wave simulation tools. We then couple this representation with heuristic optimization methods, specifically Genetic Algorithms (GA) and Surrogate-Assisted Differential Evolution for Antenna (SADEA) optimization approaches, to efficiently navigate the high-dimensional search space and identify high-performance designs aligned with target specifications. This paper describes the implementation of this workflow in MATLAB, detailing the generation of pixelized geometries, their integration into simulation routines, the formulation of fitness functions, and their integration with the optimization tools. The complete MATLAB codes, along with supporting user interface and examples, are made publicly available via GitHub to facilitate reproducibility and encourage further research in computational design of advanced planar structures.
Traditional subspace methods like SS-MUSIC often degrade under low sample support. This is further exacerbated by strong mutual coupling within antenna arrays. While deep learning offers a potential alternative, standard architectures often operate as “black boxes,” lacking the physical interpretability and mathematical rigor required for electromagnetic modeling. This paper proposes a novel Hybrid Graph Neural Network combined with a Deep Unrolling (LISTA-style) architecture. We map the Sample Covariance Matrix (SCM) onto a graph representation and utilize a shared spatial kernel (S-matrix) to maintain physical consistency across a 12-element Uniform Linear Array To resolve artifacts, we introduce a specialized “Ghost-Killer” loss function that penalizes false peaks through a quartic penalty term. The model utilizes layer-wise scalar modulators gamma (γ) and lambda (λ) to provide an interpretable “denoising curriculum.” Experimental results demonstrate that the proposed model achieves a peak success rate of 98.0% at high SNR. Crucially, the model provides super-resolution capabilities, resolving sources at a 3.0° separation with ∼70% accuracy, significantly outperforming SS-MUSIC (∼36%). Furthermore, architecture exhibits high robustness to “sample starvation,” maintaining a 78.0% success rate at only 50 snapshots, a +14% improvement over classical methods. The convergence of graph-based spatial reasoning and deep Unrolling provides a robust, interpretable framework for high-fidelity antenna array calibration and signal reconstruction, directly supporting the advancement of real-time electromagnetic diagnostics in industrial and decentralized environments.
The objective of this manuscript is to analyze and electromagnetically simulate two different horn antenna configurations, namely, a pyramidal horn antenna and a conical horn antenna. Both antennas are considered as radiating elements in a satellite communication system operating within the X-band frequency range of approximately 8–12GHz. The geometric parameters were determined analytically through mathematical calculations performed in Matlab and subsequently validated using two electromagnetic simulation environments, Ansys HFSS and CST Studio Suite. This approach enabled the evaluation of the performance of each antenna configuration as well as the assessment of their corresponding geometric structures. For the pyramidal horn antenna, fed by a WR-90 rectangular waveguide, the simulation results demonstrate good impedance matching across the X-band (S 11 = −10dB). The minimum reflection coefficient values obtained are S11 = −39.43dB at 8.22 GHz in HFSS and S 11 = −56.18dB at 8.07 GHz in CST Studio Suite. The antenna exhibits a maximum gain of 23.63dB with a narrow half-power beamwidth of approximately 6°–10°, confirming its high directivity and efficient electromagentic radiation. For the conical horn antenna, fed by a BY89 circular waveguide, the simulation results confirm good impedance matching over the 9–12 GHz frequency range, with the reflection coefficient (S 11) remaining below −10dB throughout the entire bandwidth. The minimum reflection coefficient values are S 11 = −43 dB at 11.22 GHz in HFSS and S11 = −39.70 dB in CST Studio Suite. The antenna exhibits a maximum gain of 16.43 dB and a half-power beamwidth (HPBW) of approximately 30°, which is suitable for satellite communication antennas. Circular polarization was obtained through bimodal excitation of two orthogonal TE 11 modes. An Axial Ratio AR<3dB was achieved within the main lobe region. A comparative of the two antenna configurations shows that the pyramidal horn antenna provides a higher gain and narrower beamwidth, characteristics that are suitable for applications requiring high directivity and linear polarization. The conical horn antenna demonstrates significant advantages for satellite applications due to its geometric symmetry, wider beamwidth, which is appropriate for reflector illumination, and support circular polarization through its symmetrical geometry. These characteristics make it suitable for satellite communication systems operating with circularly polarized waves. In additional, the conical horn antenna represents the basic geometric configuration from which the conical corrugated horn antenna is derived. The results obtained for this structure therefore provide a reference for the analysis and design of corrugated horn antennas in future work. The results obtained using Ansys HFSS and CST Studio Suite are in good agreement, with only minor differences observed between the two simulation environments. This consistency indicates that the antenna designs and the corresponding simulation models are suitable for X-band satellite communication applications.
Array coefficient reconstruction is a fundamental problem in antenna engineering. Although full-wave measurements enable accurate recovery of source distributions, they are often costly and experimentally complex. As an alternative, phaseless measurements provide a more practical approach; however, they lead to a nonlinear and ill-posed inverse problem due to the loss of phase information, resulting in ambiguity in reconstruction. To address this issue, a preprocessing strategy is applied to the source current distribution prior to radiation. Specifically, an additional phase-dependent term is incorporated into the original current distribution, producing a modified current source that generates more informative phaseless measurements. Based on these data, a dual-branch deep learning framework composed of two U-Net networks is developed to reconstruct the complex source distribution of two-dimensional antenna arrays. The two branches are responsible for predicting the real and imaginary components, respectively. The proposed method enables accurate recovery of both amplitude and phase information using only intensity measurements. Numerical experiments show that the model achieves a coefficient of determination (R 2) of approximately 0.98, whereas the conventional approach achieves an R 2 of approximately 0.40. The results demonstrate that the proposed preprocessing strategy significantly reduces ambiguity in phaseless inverse problems and substantially improves reconstruction accuracy. The method effectively resolves the non-uniqueness issue inherent in intensity-only measurements and outperforms conventional reconstruction approaches.
