Electrical - Electronic Engineering / Elektrik - Elektronik Mühendisliği
Permanent URI for this collectionhttps://hdl.handle.net/11147/11
Browse
36 results
Search Results
Conference Object Adaptive Limited Feedback Scheme for Stream Selection Based Interference Alignment in Heterogeneous Networks(IEEE, 2016) Beyazıt, Esra Aycan; Özbek, Berna; Le Ruyet,D.This paper presents a stream selection based interference alignment approach with imperfect channel state information for heterogeneous networks. The proposed algorithm performs the selection of a stream sequence among a predetermined set of sequences. Those selected sequences are the ones that mostly contribute to the sum rate when performing the exhaustive search. These stream sequences form a regular structure where the first stream is associated to a pico user. The effect of imperfect channel state information on the proposed algorithm is analyzed and a bit allocation scheme is proposed by deriving an upper bound on the rate loss due to quantization. © 2016 IEEE.Article Citation - WoS: 3Citation - Scopus: 4Liquid Metal-Controlled Dual-Band Doppler Radar for Enhanced Velocity Measurement(IEEE, 2024) Karatay, Anıl; Yaman, FatihDoppler radars, which are critical instruments for velocity measurement, may need to be reconfigured to adapt to different environmental conditions or for ease of use. However, conventional electrical, optical, and physical reconfiguration methods often come with several disadvantages such as deteriorated radiation pattern, reduced radiation efficiency, and high cost. Therefore, the aim of this article is to integrate microwave components that can be controlled using liquid metal (LM) displacement into a Doppler radar to adjust its main lobe direction and operating frequency to the desired values and enhance the measurement capacity of the respective radar. Through this study, multiple parameters of an operational Doppler radar have been simultaneously adjusted using LM displacement exploitation for the first time, thus avoiding the shortcomings associated with conventional reconfiguration methods. To achieve this objective, initially, a back-to-back Vivaldi antenna operating at 2.45 GHz is designed, and beam switching ability is imparted to the structure using the LM displacement method. Subsequently, various techniques are used to convert the structure into a dual-band antenna capable of simultaneous operation at 2.45 and 5.8 GHz, ensuring the desired beam switching feature at both the frequencies. In addition, a power divider capable of switching between the two operating frequencies through LM assistance is proposed, and its integration into the radar system enables the control of both main lobe direction and frequency using the proposed method.Conference Object Detection Scheme for Pnc-Based Cell-Free Mimo Systems(IEEE, 2023) Cumali, Irem; Ozbek, Berna; Kurt, Gunes KarabulutCell-free multiple-input multiple-output (cell-free MIMO) is a promising concept to overcome inter-cell interference and avoid non-uniform data rates among users by combining the best features of ultra-dense networks and MIMO. Hence, cell-free MIMO can fulfill the increasing demand on data rate with uniformly good coverage for the sixth-generation (6G) wireless communications. In addition to that, physical-layer network coding (PNC) reduces the transmission delay since it requires only two time slots instead of four time slots to exchange information between two users. In this paper, we propose a novel scheme called PNC-based cell-free MIMO to improve reliability further while reducing the transmission delay. We demonstrate the effectiveness of the proposed scheme regarding the bit error rate in different system configurations. The proposed PNC-based cell-free MIMO achieves significantly lower error probability than the conventional cell-free MIMO.Article Citation - WoS: 6Citation - Scopus: 10Using Chemosensory-Induced Eeg Signals To Identify Patients With <i>de Novo</I> Parkinson's Disease(Elsevier Sci Ltd, 2024) Olcay, Orkan; Onay, Fatih; Ozturk, Guliz Akin; Oniz, Adile; Ozgoren, Murat; Hummel, Thomas; Guducu, CagdasObjective: Parkinson's disease (PD) patients generally exhibit an olfactory loss. Hence, psychophysical or electrophysiological tests are used for diagnosis. However, these tests are susceptible to the subjects' behavioral response bias and require advanced techniques for an accurate analysis. Proposed Approach: Using well-known feature extraction methods, we characterized chemosensory-induced EEG responses of the participants to classify whether they have PD. The classification was performed for different time intervals after chemosensory stimulation to see which temporal segment better separates healthy controls and subjects with de novo PD. Results: The performances show that entropy and connectivity features discriminate effectively PD and HC participants when olfactory-induced EEG signals were used. For these methods, discrimination is over 80% for segments 100-700 and 200-800 milliseconds after stimulus onset. Comparison with Existing Methods: We compared the performance of our framework with linear predictive coding, bispectrum, wavelet entropy-based methods, and TDI score-based classification. While the entropy- and connectivity-based methods elicited the highest classification performances for olfactory stimuli, the linear predictive coding-based method elicited slightly higher performance than our framework when the trigeminal stimuli were used. Conclusion: This is one of the first studies that use chemosensory-induced EEG signals along with different feature extraction methods to classify healthy subjects and subjects with de novo PD. Our results show that entropy and functional connectivity methods unravel the chemosensory-induced neural dynamics encapsulating critical information about the subjects' olfactory performance. Furthermore, time- and frequency-resolved feature analysis is beneficial for capturing disease-affected neural patterns.Article Citation - WoS: 27Citation - Scopus: 34A New Method for Gan-Based Data Augmentation for Classes With Distinct Clusters(Pergamon-Elsevier Science Ltd, 2024) Kuntalp, Mehmet; Düzyel, OkanData augmentation is a commonly used approach for addressing the issue of limited data availability in machine learning. There are various methods available, including classical and modern techniques. However, when applying modern data augmentation methods, such as Generative Adversarial Neural Networks (GANs), to a class specific data, the resulting data can exhibit structural discrepancies. This study explores a different use of GANs as a data augmentation method that solves this problem using the electrocardiogram (ECG) signals in the MITBIH arrhythmia dataset as the example. We begin by examining the cluster structure of a specific class using tDistributed Stochastic Neighbor (t-SNE) method. Based on this cluster structure, we propose a new method for applying GANs to augment data for that class. We assess the effect of our method in a classification task using 1-D Convolutional Neural Network (CNN), Support Vector Machine (SVM), One vs one classifier (Ovo), K-Nearest Neighbors (KNN), and Random Forest as the classifiers. The results demonstrate that our proposed method could lead to better classification performance if a specific class has distinct clusters when compared to normal use of GANs.Article Citation - WoS: 5Citation - Scopus: 7Adaptive Resizer-Based Transfer Learning Framework for the Diagnosis of Breast Cancer Using Histopathology Images(Springer, 2023) Düzyel, Okan; Çatal, Mehmet Sergen; Kayan, Ceyhun Efe; Sevinç, Arda; Gümüş, AbdurrahmanBreast cancer is a major global health concern, and early and accurate diagnosis is crucial for effective treatment. Recent advancements in computer-assisted prediction models have facilitated diagnosis and prognosis using high-resolution histopathology images, which provide detailed information on cancerous tissue. However, these high-resolution images often require resizing, leading to potential data loss. In this study, we demonstrate the effect of a learnable adaptive resizer for breast cancer classification using the BreakHis dataset. Our approach incorporates the adaptive resizer with various convolutional neural network models, including VGG16, VGG19, MobileNetV2, InceptionResnetV2, DenseNet121, DenseNet201, and EfficientNetB0. Despite producing visually less appealing images, the learnable resizer effectively improves classification performance. DenseNet201, when jointly trained with the adaptive resizer, achieves the highest accuracy of 98.96% for input images of 448x448 resolution. Our experimental results demonstrate that the adaptive resizer performs better at a magnification factor of 40x compared to higher magnifications. While its effectiveness becomes less pronounced as image resolution increases to 100x, 200x, and 400x, the adaptive resizer still outperforms bilinear interpolation. In conclusion, this study highlights the potential of adaptive resizers in enhancing performance for medical image classification. By outperforming traditional image resizing methods, our work contributes to the advancement of deep neural networks in the field of breast cancer diagnostics.Article Citation - WoS: 2Citation - Scopus: 2A Non-Resonant Approach for Dielectric Constant Reconstructions Via Newton Iterations(Elsevier, 2023) Özkal, Ceren; Yaman, FatihIn this study, a Newton–Raphson-based iterative method has been proposed to obtain dielectric constants accurately from measurements. The originalities of the approach lie in its applicability at non-resonant frequencies, which brings a significant experimental simplicity by avoiding critical coupling, expansion of available frequencies in different bands with the same cost-efficient low-Q (?60) cavity. The direct problem involves either measuring power values inside a cavity (14.6 × 5 × 20.6) cm via a spectrum analyzer or simulating the complete setup via CST-MWS software at one of the non-resonant modes, 1.5 GHz. The solution to the inverse problem provides fastly converging results with an error rate of 1% for the unknown permittivities. The experiments were carried out using five different liquid samples even though the proposed technique does not have a limitation on solid materials. Applicability and the effectiveness of the introduced method is illustrated in detail and comparisons with the perturbation method is provided. © 2023 Elsevier GmbHArticle Citation - WoS: 6Citation - Scopus: 7Time-Resolved Eeg Signal Analysis for Motor Imagery Activity Recognition(Elsevier, 2023) Olcay, Bilal Orkan; Karaçalı, BilgeAccurately characterizing brain activity requires detailed feature analysis in the temporal, spatial, and spectral domains. While previous research has proposed various spatial and spectral feature extraction methods to distinguish between different cognitive tasks, temporal feature analysis for each separate brain region and frequency band has been largely overlooked. This study introduces two novel approaches for recognizing cognitive activity: temporal entropic profiling and time-aligned common spatio-spectral patterns analysis. These approaches capture and use discriminative short-lived signal segments for motor imagery activity recognition. In Approach-1, we evaluated nine different measures to determine timing parameters that showed altered behavior associated with maximal inter-activity differences, which we then used in a machine-learning framework. In Approach-2, we used the best-performing signal characteristic measures from Approach-1 to determine the optimum latency of each channel at each frequency band for a CSP-based activity recognition strategy. We evaluated both approaches on two online available motor imagery EEG datasets and achieved average recognition accuracy levels of 86%. We compared our methods with four established BCI methods. The performance results show that our approaches exceeded the benchmark methods' performances, with notable improvements in the proposed time-aligned common spatio-spectral patterns approach. This study demonstrates that motor imagery recognition performance is improved when a temporal analysis is adopted alongside spatio-spectral neural feature analysis and that timing parameters associated with the maximal entropic difference of EEG segments to the cognitive tasks varied between different brain regions and subjects. © 2023 Elsevier LtdConference Object Citation - Scopus: 1Interference Mitigation for Device-To Based Cellular Communications(IEEE, 2022) Acar, Süleyman Onur; Özbek, BernaDevice-to-device (D2D) communication underlaying cellular networks can improve the performance of cellular systems and it provides an effective way to meet growing mobile traffic and capacity demand. When user equipments are located in close proximity, they can communicate through direct links. In this case, D2D links can increase both energy and spectrum efficiency by reusing uplink (UL) cellular resources while satisfying the users' quality-of-service requirements. However, integrating D2D links into the cellular infrastructure causes an interference since D2D communication can increase co-channel interference and degrade cellular users' transmission link quality. In this paper, the interference mitigation techniques including power control, multiple antenna and resource allocation based on graph coloring are proposed for D2D communications underlaying cellular systems to increase the data rate of both the cellular users and D2D pairs. Compared to the prior works, in the proposed algorithm, D2D and cellular users have same priority for resource allocation. Finally, the proposed algorithm improves the overall system capacity significantly.Article Citation - WoS: 2Citation - Scopus: 2Deep Learning Based Adaptive Bit Allocation for Heterogeneous Interference Channels(Elsevier, 2021) Aycan, Esra; Özbek, Berna; Le Ruyet, DidierThis paper proposes an adaptive bit allocation scheme by using a fully connected (FC) deep neural network (DNN) considering imperfect channel state information (CSI) for heterogeneous networks. Achieving an accurate CSI has a crucial role on the system performance of the heterogeneous networks. Different quantization techniques have been employed to reduce the feedback overhead. However, the system performance cannot increase linearly with the number of bits increasing exponentially. Since optimizing the total number of bits is too complex for the entire network, an initial step is performed to distribute the bits to each cell in the conventional method. Then, the distributed bits are further allocated to each channel optimally. In order to enable direct allocation for the entire network, a FC-DNN based method is presented in this study. The optimized number of bits can be directly obtained for a different number of bits and scenarios by the proposed approach. The simulations are performed by using various scenarios with different allocation schemes. The performance results show that the DNN based method achieves a closer performance to the conventional approach. (C) 2021 Elsevier B.V. All rights reserved.