Ilgin, Fatih Yavuz; Çelik, Husamettin

In this paper, we propose a novel spectrum detection method for cognitive radios using covariance matrix eigenvalues and its trace. Unlike some traditional eigenvalue and covariance-based detectors, the proposed method is independent of noise power and computationally simple. An optimal threshold calculation approach is also proposed based on the statistical characteristics of the test statistic. This threshold optimization improves detection reliability under different signal-to-noise ratio (SNR) conditions. This common usage improves detection reliability, especially under low SNR conditions and limited sample sizes. Theoretical analyses are provided in detail to characterize the statistical behavior of the proposed test statistic for both hypotheses. Detailed simulation results show that the proposed method is advantageous in terms of computational cost and detection performance.

Karakose, Perihan; Ince, Elif Gokce; Eren, Erkin; Atar, Muhammed; Cakil, Burak; Ince, Ozan

Earthquake-resistant design of reinforced concrete (RC) buildings is initiated during the architectural design stage, where the placement and directions of vertical load-bearing elements are defined. The configuration of these elements plays a critical role in the seismic performance of RC structures. In this study, a decision-support system is proposed to automatically identify load-bearing elements from architectural floor plans and to assess torsional irregularity and shear wall distribution. A dataset comprising 600 architectural floor plans of existing buildings was compiled, in which structural components were manually annotated and used to train a YOLO (You Only Look Once)-based detection model. The developed system detects vertical load bearing elements in the floor plan and subsequently evaluates their arrangement using structural engineering criteria, including torsional irregularity and shear wall ratio. The results show that the proposed approach enables rapid and accurate evaluation of architectural floor plans, offering effective support during the early design phase.

Ozden, M.; Coban, O.; Karacali, T.

This study investigates the structural, morphological, and optical properties of PZT (lead zirconate titanate) thin films deposited on Al2O3 substrates via RF sputtering, with a particular focus on the effects of annealing temperature. The films were annealed at 500°C, 600°C, and 700°C, and characterized using XRD, SEM (including cross-sectional imaging), AFM, EDX, and UV–VIS spectroscopy. SEM and AFM images were additionally analyzed using MATLAB-based image processing techniques. XRD analysis revealed a dominant pyrochlore phase at 500°C, whereas the perovskite phase became prominent at higher temperatures. Grain sizes were estimated using the Scherrer equation and further validated through quantitative image analysis. Grain size evaluations based on three complementary techniques—XRD (via the Scherrer equation), MATLAB-assisted SEM image analysis, and AFM topography—consistently demonstrated a particle growth trend with increasing annealing temperature. While XRD yielded average crystallite sizes from diffraction peak broadening, image-based particle analysis and surface profiling provided direct morphological confirmation. The agreement among these independent methods reinforces the conclusion that higher annealing temperatures promote grain coarsening, likely driven by enhanced atomic mobility and crystallization dynamics. SEM images confirmed the formation of homogeneous and crack-free films, while cross-sectional SEM indicated a decrease in film thickness with increasing temperature. EDX analysis verified the elemental composition with only minor deviations. Tauc plots, based on a direct allowed transition, showed that the optical band gap decreased from 3.5 eV in the pyrochlore phase (500°C) to approximately 3.35 eV in the perovskite phase (600°C and 700°C), reflecting improved crystallinity and reduced structural disorder. Overall, the results highlight the critical role of annealing temperature in tailoring the crystallinity, surface morphology, and optical properties of PZT thin films, offering valuable insights for their application in high-performance electronic and optoelectronic devices.

Karakas, Betul; Kulluk, Sinem

In complex industrial manufacturing processes, various defects may occur in products because of the poor quality of the equipment, materials, and human factors. These defects may lead to undesirable situations due to reasons such as increasing costs, shortening service life, reducing customer satisfaction, and adversely affecting quality. To reduce these problems, it is crucial to detect products with such defects quickly and effectively. Therefore, within the scope of this study, a hybrid Convolutional Neural Network (CNN)-Support Vector Machine (SVM) hybrid model is introduced to improve the classification accuracy of production defects. The proposed framework incorporates a CNN for feature extraction, coupled with a Least Absolute Shrinkage and Selection Operator (LASSO) to facilitate feature selection. This approach aims to mitigate the dimensionality of the feature space and is complemented by the implementation of a SVM classifier. The performance of the model was tested on three widely used publicly available datasets: metal casting, textile fabric, and food images. Significant and high-performance results in both classification accuracy and execution time, which are important performance metrics for real-time applications, were obtained by the proposed hybrid CNN-SVM algorithm. Statistical analyses are also performed to test whether there is a significant difference between the CNN-SVM hybrid and conventional CNN and between CNN-SVM hybrid and the previous studies in literature. This study demonstrates that the AlexNet-SVM hybrid with LASSO outperforms the conventional CNN in terms of execution time. Furthermore, the AlexNet-SVM hybrid without LASSO achieved classification accuracies of 98,29%, 99,82%, and 97,37% for the casting, food, and fabric datasets, respectively.

Arslantas, Hatice; Okdem, Selcuk

Nowadays, studies on indoor localization systems based on wireless systems are increasing widely. Indoor localization is the process of determining the location of objects or people inside a building. Global Navigation Satellite System (GPS) signals do not provide sufficient location data indoors because they are interrupted or completely lost in closed areas. For this reason, studies on indoor localization system design with machine learning and deep learning techniques based on Wi-Fi technology are increasing. In this study, we propose a method and training strategy that is entirely based on a Convolutional Neural Network (CNN) and a combined autoencoder that automatically extracts features from Wi-Fi fingerprint samples. In this model, we coupled an autoencoder and a CNN and we trained them simultaneously. Thus, we guarantee that the encoder and the CNN are trained simultaneously. The proposed system was evaluated on the UJIIndoorLoc and Tampere datasets. The experimental results show that the proposed model performs significantly better than the current state-of-the-art methods in terms of location coordinates (x, y) localization. In our study, runtime analysis is also presented to show the real-time performance of the network we proposed.

Alici-Karaca, Demet; Akay, Bahriye

Prostate cancer is one of the most widespread and mortal cancers among men worldwide. Pathological evaluation is essential for the definitive diagnosis of prostate cancer, the symptoms of which are not evident during the early disease. During the pathological evaluation to determine the severity of cancer, Gleason grading, which is difficult to assign, is time-consuming, and the result may vary depending on the pathologist's experience. In order to make a fast and objective evaluation, deep learning systems have been recently used to automate prostate cancer diagnosis and grading. The inherent difficulties of detecting Gleason grades can be overcome with deep learning architectures, which present fast and precise classification rates. The performance of the methods proposed for Gleason grading in the literature is low, especially in multi-class problems, due to their model structures. Some mechanisms for better learning of critical representations are needed to determine the Gleason grading accurately. Therefore, in this study, a novel deep learning model called Eff4-Attn is proposed for detecting and grading prostate cancer. In this model, combining the Efficient Channel Attention (ECA) mechanism with EfficientNet-B4 distinguishes the tissues in the prostate cancer pathology images much better. In addition, the effect of data distribution in classes in the public DiagSet dataset on performance was investigated. Experimental studies have been carried out to perform three classification tasks for various magnification ratios: 8-class classification, 4-class classification, and 2-class classification for diagnosing prostate cancer and determining cancer severity, and promising results have been obtained. At 40x magnification ratio, the best result was achieved with 96.18% test accuracy in the 2-class classification task, that is, in detecting the presence of cancer. In determining the severity of cancer, at a 40x magnification ratio, a test accuracy of 94.86% was obtained. The proposed deep and light model has considerable potential to assist pathologists in detecting the characteristics of Gleason grades.

Cavli, Nazlican; Ozkop, Emre

This study investigates the performance of a single-phase 5-level H-Bridge Neutral Point Clamped (HBNPC) inverter across various operating conditions. These conditions encompass variations in output power, frequency, modulation index, and modulation techniques. To adapt modulation strategies typically employed in three-phase inverter structures, we apply them to a single-phase HBNPC inverter. The modulation strategies under consideration include level-shifted (LS) based on phase disposition pulse width modulation (PD-PWM), third harmonic injection pulse width modulation (THI-PWM), space vector pulse width modulation (SV-PWM), and modified PWM (MPWM). These strategies are developed using different types of reference signals (RS) and carrier signals (CS). Initially, we analyze the inverter's output voltage and current waveforms across various modulation strategy states. Subsequently, we explore the relationships between efficiency, total harmonic distortion (THD), and variable output power across different modulation strategy states. We uncover distinct patterns between modulation strategy-THD, efficiency, and output voltage for maximum inverter output power. Furthermore, we examine the relationship between the modulation index and THD for each modulation strategy state, considering different switching frequencies. Additionally, we investigate the relationship between switching frequency and THD across different modulation index values, approaching the analysis from multiple perspectives.

Gebesoglu, Durmus; Ertay, Agah Oktay; Simsek, Serkan

Electromagnetic Bandgap (EBG) structures that exhibit various performances, such as preventing the electromagnetic wave propagation and reflecting an incident wave within the stopband, have unique electromagnetic characterization. Due to this reason, accurate analysis and design of the EBG structures are crucial to enhance the integrated system performance. This paper concentrates on the characterization of some planar EBGs using the Auxiliary Functions of Generalized Scattering Matrix (AFGSM) methods, with particular importance on an in- depth consideration of its bandgaps. The AFGSM method is applied to the planar EBG structures in the literature for the first time. The well-known kinds (symmetric and asymmetric cases) of mushroom type and multilayer EBG structures are considered to verify the presented method. Analysis results are compared with the Conventional Eigenvalue Equation (C-EIV) and the Generalized Scattering Matrix based Eigenvalue Equation (GSM-EIV) methods. Low computation load and accurate results are obtained to analyze the planar EBG structures with the AFGSM method due to using transmission line model. In addition, a design methodology is proposed for a chosen planar EBG structure using the AFGSM methods. Geometrical parameters of interested EBG problems are determined for acquiring the stopband frequency region of interest using the scattering parameters of unit cell configuration. The mushroom EBG model along one and two-dimensional axes is used in an antenna application to decrease mutual coupling between antenna elements. Three different scenarios are simulated in the HFSS electromagnetic simulation design environment to understand the effect of mutual coupling reduction in the antenna problem of the designed EBG structure via the AFGSM method. All designed antennas are manufactured, and the measurement results are in good agreement with the simulation results. The measurement results of the fabricated antenna application example including designed EBG using the proposed AFGSM method are compared with the existing similar problems with the same and different EBG models. It has been demonstrated that bandgap analysis, design of the planar EBG structures and integration of considered EBG model to a design application can be accurately and quickly achieved with the given methodology using the AFGSM method.

Gelen, Aykut G.; Atasoy, Ayten

The SLAM problem for autonomous robots can be greatly improved by using event-based cameras. Compared to others, event-based cameras consume very low power while providing great temporal resolution and dynamic range. In this study, we propose a convolutional neural SLAM framework based solely on the event data. Event-based cameras generate events for pixels whose brightness changes. Therefore, the event data is rich in motion and edge information. The purpose of the proposed framework is to make all estimations using encoded information in event data. The proposed solution is in the form of keyframe-based visual SLAM, consisting of three neural networks that can estimate the relative camera pose, log-depth and features for loop closure detection. In the study, network architectures and learning curves for the trained networks are presented and it is shown that networks can learn the problems successfully. The proposed method has been developed and tested on a new dataset generated by the CARLA simulator. It has been shown that the proposed method is a SLAM solution and it can keep global drift under control with loop closure estimations. Evaluation metrics for estimations, evaluation of the global model and an analysis of run-time performance are also presented.

Orman, Kamil

In this study, a memristor-based 2-DOF PI controller (2-DOF Mem-PI) was designed for the temperature profile tracking control of a model of heat flow experiment (HFE) setup. A simulation study is presented in which the performance of the designed controller is compared with a standard 2-DOF PI controller. Compared to 1-DOF control structures, 2-DOF controllers that include an extra adjustable parameter perform better in terms of response to disturbances and improving the transient response of the system. In addition, memristor-based controllers (Mem-PI and 2-DOF Mem-PI) and standard controllers (PI and 2-DOF PI) were compared and it was determined that because of the variable memristance value, the control structures containing memristors showed an adaptive feature. The simulation results demonstrated the success of the proposed controller in temperature profile reference tracking and showed the memristor to be applicable in nonlinear control structures.