arXiv subjects

Okan Yurduseven

Publications and source records attributed to Okan Yurduseven.

At least 19 recordsLinked to original sources

arXiv2026

Generative Adversarial Reconstruction with Adaptive Thresholding for Obstructed Targets in Computational Microwave Imaging

In this work, an end-to-end generative adversarial framework for computational microwave imaging (CMI) is proposed to reconstruct the targets of interest directly from the measurements of targets obstructed by undesired objects. It integrates a conditional generative adversarial network (cGAN) with a learnable soft-threshold module (STM) to adaptively suppress non-target related information. The proposed framework is evaluated on a diverse dataset comprising MNIST digits obstructed by E-MNIST letters for training and testing, as well as on measurements acquired with an experimental CMI system. In addition, further studies involving objects with different geometries are conducted, demonstrating that the proposed approach can be adapted to other types of objects. Numerical experiments show that the proposed cGAN-STM achieves a normalized mean square error (NMSE) of 0.066 and a structural similarity index (SSIM) of 0.876 under ideal conditions. Comprehensive analyses, including benchmarking, analysis of the STM mechanism, and evaluation under different obstruction sizes, are also conducted. The performance of the model under different signal-to-noise ratio (SNR) scenarios is also evaluated, achieving reasonable reconstruction quality at 15 dB SNR with an NMSE of 0.168 and an SSIM of 0.700. Even at low SNR levels, recognizable target outlines are preserved. These results highlight the effectiveness and adaptability of the proposed method.

eess.SP↗

arXiv2026

Computational Microwave Imaging Relying on Orbital Angular Momentum Transmitarrays for Improved Diversity

This work proposes the use of orbital angular momentum (OAM) waves to improve the performance of a computational imaging (CI) system. Specifically, in contrast to a solely frequency-diverse operation, leveraging multiple OAM waves leads to a significant increase in the diversity of the measurement modes of a CI system. This significantly reduces the frequency bandwidth required to achieve high-quality image reconstructions. A proof-of-concept prototype working at Ka-band frequencies is used to validate the proposed approach. The prototype consists of two metalized three-dimensional (3D) printed cavities, with fully-dielectric transmitarrays inside that generate OAM waves. Imaging results from various targets reveal that the CI system achieves superior imaging quality when multiple OAM waves are considered, compared to when it solely relies on frequency-diversity. This is specially noticeable in the case of complex distributed targets, which can only be reconstructed with the prototype when multiple OAM waves are used. Furthermore, it is shown that accurate image reconstructions can be obtained employing only one eighth of the operational bandwidth of the frequency-diverse system.

physics.ins-det↗

arXiv2026

ML-Assisted Bulk Resource Allocation: Custom Outage-Based Loss Function and Reliability Analysis

Machine learning (ML)-assisted outage-based resource allocation has recently emerged as an effective alternative to conventional scheduling methods in reliability-critical wireless systems. However, existing approaches are fundamentally limited to single-resource allocation, whereas modern and emerging systems increasingly require the simultaneous allocation of multiple resources to meet aggregate rate and reliability constraints. In this paper, we extend outage-based learning to the bulk resource allocation regime, where a user requires at least $D$ reliable resources from a pool of $R$ candidates. We first introduce a practical allocation policy, termed gate + top-$D$ allocation (GTBA), which combines threshold-based admission control with ranking-based selection. We then propose a novel ranking-aware bulk outage loss (RBOL) that provides a differentiable surrogate for the bulk outage event induced by GTBA, explicitly accounting for both gate failures and ranking errors near the selection boundary. An exact reliability analysis is developed, establishing a decomposition of bulk outage probability (BOP), identifying dominant failure mechanisms and deriving an oracle lower bound that characterizes the fundamental performance limit. Extensive simulations under balanced, light and heavy stress regimes demonstrate that RBOL consistently outperforms conventional pointwise losses and baselines, achieving substantial reductions in BOP and remaining significantly closer to the oracle bound across a wide range of operating conditions. These results confirm that set-level ranking-aware training objectives are essential for reliable ML-assisted bulk resource allocation.

eess.SP↗

arXiv2025

Design of Rectangular Waveguide-fed Metasurfaces for Near-Field Shaping using a Coupled Dipole Model

We present the design of rectangular waveguide-excited metasurfaces for near-field shaping using a coupled dipole framework. Waveguide-fed metasurfaces are array-like radiating systems typically constructed from one or more waveguides loaded with a series of subwavelength metamaterial apertures that function as radiators. The use of subwavelength radiating elements distributed across the aperture enables electromagnetic field control with subwavelength precision, offering significant potential for near-field shaping. Leveraging these capabilities, we demonstrate that the near-field patterns of rectangular waveguide-fed metasurfaces can be tailored using the coupled dipole model, which accounts for mutual interactions between metamaterial radiating elements. The validity and effectiveness of the proposed approach are verified through full-wave simulations and experiments in the X-band.

physics.optics↗

arXiv2025

Adversarial Learning-Based Radio Map Reconstruction for Fingerprinting Localization

This letter presents a feature-guided adversarial framework, namely ComGAN, which is designed to reconstruct an incomplete fingerprint database by inferring missing received signal strength (RSS) values at unmeasured reference points (RPs). An auxiliary subnetwork is integrated into a conditional generative adversarial network (cGAN) to enable spatial feature learning. An optimization method is then developed to refine the RSS predictions by aggregating multiple prediction sets, achieving an improved localization performance. Experimental results demonstrate that the proposed scheme achieves a root mean squared error (RMSE) comparable to the ground-truth measurements while outperforming state-of-the-art reconstruction methods. When the reconstructed fingerprint is combined with measured data for training, the fingerprinting localization achieves accuracy comparable to models trained on fully measured datasets.

eess.SP↗

arXiv2025

FinGAN: An Interpretable RSS Generation Network for Scalable Fingerprint Localization

This work introduces FinGAN, a robust received signal strength (RSS) data generator designed to expand RSS fingerprint datasets. Compared to existing generative adversarial models that either rely on known reference positions (RPs) or depend on predefined priors, FinGAN learns the latent information between RPs and RSS values by maximizing the mutual information between the generated RSS data and the RPs, enabling an end-to-end RSS generation directly from RPs. This allows us to accurately generate RSS data for previously unmeasured RPs. Both quantitative and qualitative evaluations demonstrate that FinGAN produces synthetic RSS data closely aligned with real RSS sample collected from the on-site experiment, preserving localization performance comparable to that achieved with complete real-world datasets. To further validate its generalizability, FinGAN is also trained and evaluated on open-source datasets from three typical office environments,and the results demonstrate consistent performance across different scenarios.

eess.SP↗

arXiv2025

Integrated Image Reconstruction and Target Recognition based on Deep Learning Technique

Computational microwave imaging (CMI) has gained attention as an alternative technique for conventional microwave imaging techniques, addressing their limitations such as hardware-intensive physical layer and slow data collection acquisition speed to name a few. Despite these advantages, CMI still encounters notable computational bottlenecks, especially during the image reconstruction stage. In this setting, both image recovery and object classification present significant processing demands. To address these challenges, our previous work introduced ClassiGAN, which is a generative deep learning model designed to simultaneously reconstruct images and classify targets using only back-scattered signals. In this study, we build upon that framework by incorporating attention gate modules into ClassiGAN. These modules are intended to refine feature extraction and improve the identification of relevant information. By dynamically focusing on important features and suppressing irrelevant ones, the attention mechanism enhances the overall model performance. The proposed architecture, named Att-ClassiGAN, significantly reduces the reconstruction time compared to traditional CMI approaches. Furthermore, it outperforms current advanced methods, delivering improved Normalized Mean Squared Error (NMSE), higher Structural Similarity Index (SSIM), and better classification outcomes for the reconstructed targets.

eess.SP↗

arXiv2025

Metasurfaces-Enabled Wave Computing for Future Wireless Systems: Opportunities and Challenges

The next generations of wireless networks are envisioned to integrate communications, sensing, and computing into a unified platform, demanding ultra-high data rates, submillisecond latency, and unprecedented energy efficiency. However, conventional digital processors face limitations in scalability, cost, and power consumption that hinder this vision. Wave computing, enabled by programmable metasurfaces, offers an alternative paradigm according to which signal processing operations are implemented in the domain of the propagation of electromagnetic waves. This approach transforms metasurfaces from passive wavefront shapers into functional analog processors capable of executing tasks such as beamforming, sensing, imaging, and machine learning at the speed of light with minimal power consumption. This article provides an overview of metasurface-enabled wave computing, highlighting its fundamental principles and key application scenarios for future wireless systems, including integrated sensing and communications, artificial intelligence acceleration, over-the-air channel estimation, and computational electromagnetic imaging. Future research directions are outlined in response to the major open challenges of the technology, aiming to enable large-scale deployment of wave computing in practical wireless networks.

eess.SP↗

arXiv2024

Differential Evolution-Based End-Fire Realized Gain Optimization of Active and Parasitic Arrays

We propose a novel approach for boosting the realized gain in enhanced directivity arrays with both active and parasitic dipoles as radiating elements. The optimization process involves two main objectives: maximizing the end-fire gain and minimizing the reflection coefficient to ensure high realized gain. In the first step, the current excitation vector of the fully driven array is selected to maximize the end-fire gain. Then, all but one of the dipoles are reactively loaded according to their input impedance. Following that, the optimization focuses on the inter-element distance, computing the one that offers a favorable balance between the gain and the total efficiency. This multi-objective optimization leverages the differential evolution (DE) algorithm and utilizes a simple wire dipole as the unit element. Full-wave simulations further confirm the accuracy of our theoretical results. Our two- and three-element parasitic arrays achieve realized gain comparable to state-of-the-art designs, without relying on intricate unit elements or resource-intensive simulations. Moreover, our four- and five-element parasitic arrays deliver the highest realized gain values reported in the literature. The simplicity of our approach is validated by significant time savings, with theoretical models completing optimizations much faster than full-wave simulations. Additionally, a sensitivity analysis confirms the robustness of the proposed optimization algorithm, demonstrating that the optimized design parameters remain effective even under small deviations in loads and element positions. Finally, the proposed parasitic arrays are well-suited for base station antennas due to their compact design, reduced power consumption, and simplified hardware requirements, making them ideal for modern communication systems.

eess.SP↗

arXiv2024

Near-Field Localization with Antenna Arrays in the Presence of Direction-Dependent Mutual Coupling

Localizing near-field sources considering practical arrays is a recent challenging topic for next generation wireless communication systems. Practical antenna array apertures with closely spaced elements exhibit direction-dependent mutual coupling (MC), which can significantly degrade the performance localization techniques. A conventional method for near-field localization in the presence of MC is the three-dimensional (3D) multiple signal classification technique, which, however, suffers from extremely high computational complexity. Recently, two-dimensional (2D) search alternatives have been presented, exhibiting increased complexity still for direction-dependent MC scenarios. In this paper, we devise a low complexity one-dimensional (1D) iterative method based on an oblique projection operator (IMOP) that estimates direction-dependent MC and the locations of multiple near-field sources. The proposed method first estimates the initial direction of arrival (DOA) and MC using the approximate wavefront model, and then, estimates the initial range of one near-field source using the exact wavefront model. Afterwards, at each iteration, the oblique projection operator is used to isolate components associated with one source from those of other sources. The DOA and range of this one source are estimated using the exact wavefront model and 1D searches. Finally, the direction-dependent MC is estimated for each pair of the estimated DOA and range. The performance of the proposed near-field localization approach is comprehensively investigated and verified using both a full-wave electromagnetic solver and synthetic simulations. It is showcased that our IMOP scheme performs almost similarly to a state-of-the-art approach but with a 42 times less computational complexity.

eess.SP↗

arXiv2024

Near-Field Localization with an Exact Propagation Model in Presence of Mutual Coupling

Localizing near-field sources considering practical arrays is important in wireless communications. Array-based apertures exhibit mutual coupling between the array elements, which can significantly degrade the performance of the localization method. In this paper, we propose two methods to localize near-field sources by direction of arrival (DOA) and range estimations in the presence of mutual coupling. The first method utilizes a two-dimensional search to estimate DOA and the range of the source. Therefore, it suffers from a high computational load. The second method reduces the two-dimensional search to one-dimensional, thus decreasing the computational complexity while offering similar DOA and range estimation performance. Besides, our second method reduces computational time by over 50% compared to the multiple signal classification (MUSIC) algorithm.

eess.SP↗

arXiv2024

Optimization of Super-Directive Linear Arrays with Differential Evolution for High Realized Gain

Due to the low impedance and high feeding currents, it is naturally challenging to design super-directive antenna arrays that perfectly match the feed line, and this becomes almost impossible as the number of elements increases. In this paper, we assert that it is crucial to consider the trade-off between directivity and overall efficiency (to achieve high realized gain) before employing super-directive arrays in real-world applications. Given this trade-off (high directivity and low mismatch for high realized gain), a 4-element dipole array (unit array) is optimized using the differential evolution (DE) algorithm. Then, the performance of the unit array in subarray configuration scenarios is analyzed. Finally, the obtained parameters are verified using the CST full-wave simulation software. The results clearly indicate that the proposed unit array is a strong candidate for dense array applications, particularly in the context of massive multiple-input multiple-output (MIMO), thanks to its notable high gain and efficiency.

eess.SP↗

arXiv2024

Super-Directive Antenna Arrays: How Many Elements Do We Need?

Super-directive antenna arrays have faced challenges in achieving high realized gains ever since their introduction in the academic literature. The primary challenges are high impedance mismatches and resistive losses, which become increasingly more dominant as the number of elements increases. Consequently, a critical limitation arises in determining the maximum number of elements that should be utilized to achieve super-directivity, particularly within dense array configurations. This paper addresses precisely this issue through an optimization study to design a super-directive antenna array with a maximum number of elements. An iterative approach is employed to increase the array of elements while sustaining a satisfactory realized gain using the differential evolution (DE) algorithm. Thus, it is observed that super-directivity can be obtained in an array with a maximum of five elements. Our results indicate that the obtained unit array has a $67.20\%$ higher realized gain than a uniform linear array with conventional excitation. For these reasons, these results make the proposed architecture a strong candidate for applications that require densely packed arrays, particularly in the context of massive multiple-input multiple-output (MIMO).

eess.SP↗

arXiv2022

Hollow Rectangular Waveguide-fed Holographic Beamforming Antenna Additively Manufactured (3D Printed) with Conductive Polymer

We present the design and fabrication of 3D printed holographic beamforming antennas. The antennas utilize additively manufactured hollow rectangular waveguides that feed radiating rectilinear slots inserted into the upper conducting wall. The lengths of the individual slots are altered to implement a holographic beamforming solution designed using a coupled dipole formalism. For rapid verification, the designed antennas are fabricated using a desktop dual-extrusion fused filament 3D printer. The body of each antenna and its inner conducting surface are respectively printed using polylactic acid and biodegradable conductive polyester composite material (i.e., Electrifi), which is later deposited with a layer of copper on its surface to improve surface conductivity and reduce surface roughness. The beamforming performance of the fabricated antennas is confirmed via experiments. The 3D printed metasurface antennas using the proposed fabrication technique illustrate emerging capabilities in the rapid prototyping of complex electromagnetic structures.

physics.app-ph↗

arXiv2022

Optimizing Polarizability Distributions for Metasurface Apertures with Lorentzian-Constrained Radiators

We present a design strategy for selecting the effective polarizability distribution for a metasurface aperture needed to form a desired radiation pattern. A metasurface aperture consists of an array of subwavelength metamaterial elements, each of which can be conceptualized as a radiating, polarizable dipole. An ideal polarizability distribution can be determined by using a holographic approach to first obtain the necessary aperture fields, which can then be converted to a polarizability distribution using equivalence principles. To achieve this ideal distribution, the polarizability of each element would need to have unconstrained magnitude and phase; however, for a single, passive, metamaterial resonator the magnitude and phase of the effective polarizability are inextricably linked through the properties of the Lorentzian resonance, with the range of phase values restricted to a span of at most 180 degrees. Here, we introduce a family of mappings from the ideal to the available polarizability distributions, easily visualized by plotting both polarizabilities in the complex plane. Using one of these mappings it is possible to achieve highly optimized beam patterns from a metasurface antenna, despite the inherent resonator limitations. We introduce the mapping technique and provide several specific examples, with numerical simulations used to confirm the design approach.

physics.app-ph↗

arXiv2022

Spatio-temporal analysis of electromagnetic field coherence in complex media

We study the coherence in time and space of electromagnetic fields propagated through complex media. Whether for localization, imaging or telecommunication, the development of dedicated numerical techniques is generally based on the exploitation of simplified models considering either coherent or diffuse fields. The optimization of such applications in conditions of partial coherence can therefore be particularly challenging, requiring the development of hybrid algorithms adaptable to prior knowledge on the processed fields. The objective of this work is to provide numerical techniques for decomposing an electromagnetic field into subspaces that can then be filtered according to their level of spatial and temporal coherence. In contrast to the studies carried out on space-space transfer matrices notably used for the calculation of Wigner-Smith operators, these decompositions are carried out on space-time matrices in order to facilitate the study of temporal dispersion. The theory is developed for illustrative purposes using experimental results from a leaky resonant system but seem to be applicable to any scattering and reverberating media capable of transforming localized and coherent excitations into complex and diffuse distributions. To conclude this work, the proposed technique is exploited to improve image reconstruction in a millimeter-wave computational imaging demonstration. In the studied context and from a more general perspective, we propose a technique to select the most suitable subspaces for each application operating under conditions of partial coherence, whether these correspond in the most extreme cases to ballistic paths or to diffuse fields.

physics.comp-ph↗

arXiv2021

Smart Radio Environments

This Roadmap takes the reader on a journey through the research in electromagnetic wave propagation control via reconfigurable intelligent surfaces. Meta-surface modelling and design methods are reviewed along with physical realisation techniques. Several wireless applications are discussed, including beam-forming, focusing, imaging, localisation, and sensing, some rooted in novel architectures for future mobile communications networks towards 6G.

physics.class-ph↗

arXiv2021

Graph Attention Network Based Single-Pixel Compressive Direction of Arrival Estimation

In this paper, we present a single-pixel compressive direction of arrival (DoA) estimation technique leveraging a graph attention network (GAT)-based deep-learning framework. The physical layer compression is achieved using a coded-aperture technique, probing the spectrum of far-field sources that are incident on the aperture using a set of spatio-temporally incoherent modes. This information is then encoded and compressed into the channel of the coded-aperture. The coded-aperture is based on a metasurface antenna design and it works as a receiver, exhibiting a single-channel and replacing the conventional multichannel raster scan-based solutions for DoA estimation. The GAT network enables the compressive DoA estimation framework to learn the DoA information directly from the measurements acquired using the coded-aperture. This step eliminates the need for an additional reconstruction step and significantly simplifies the processing layer to achieve DoA estimation. We show that the presented GAT integrated single-pixel radar framework can retrieve high fidelity DoA information even under relatively low signal-to-noise ratio (SNR) levels.

eess.SP↗