Theses supervised by Prof. Dr. Oğuz Bayat
21 theses · Altınbaş University, Yeditepe University
Multi-approached histopathology images classification
Histological images play a crucial role in diagnosing diseases, especially breast cancer, which remains a major health concern for women, and colon and lung for people worldwide. Computer-aided diagnosis tools significantly assist physicians in early detection and treatment planning, helping reduce mortality rates. Although, Convolutional neural networks (CNNs) based on deep learning have proven effective in distinguishing benign from malignant breast cancers, the scarcity of variability and feature diversity in small datasets would steer clear of CNNs and result in overfitting instances, as well as diminished discrimination. Deep learning is very sensitive to the hyperparameter optimization, and for the reason, the learning hierarchies could be the solution for network's comprehension of the relationship subtlety between the of the data types. In this context, this dissertation introduced a HAFMAB-Net: Hierarchical Adaptive Fusion based on Multilevel Attention-Enhanced Bottleneck Neural Network. The network comprises two pathways utilizing an enhanced Bottleneck architecture with attention mechanisms to extract both global and spatial features. It incorporates a Deeper Spatial Attention Aggregator Module to boost the representation of locative features by focusing on key spatial regions, improving the discriminative power of aggregated features. Additionally, a modified Adaptive Fusion Module combines the enhanced global and boosted spatial features into a comprehensive and enriched feature representation, which is subsequently used for cancer classification based on whole image. The histopathological image could include more than one type cancer inside one image, to address this issue, we proposed a novel MATERB Net: Multiscale Attention transformation based on Enhanced Residual Block for Patch-level breast histopathology images prediction. The proposed framework employs extraction feature processes based on multiscale to capture more complementary global and local characteristics to enhance the learning process of the model in coping with morphological variability within the tissue. In Additional the model uses Attention transformation techniques based on self-attention and cross-attention is used to learn the relationship between the pixels of the image and focus on the meaningful details and information in calculating the attention weights. Self-attention works on capturing and learning the relation between the extracted features based on multi-magnification factors sub-images, providing various levels of details and information, which feed into the cross-attention beside the extracted features from full images. Cross-attention works to provide meaningful and context-aware representation integrated with the global context and details of each label. Moreover, the proposed net used a Balanced Focal Loss function, addressing the treating all misclassified samples equally, and preventing the model from memorizing easy samples to reduce overfitting.
Propose an architecture for e-health within smart cities
According to the World Health Organization: eHealth is the cost-effective and secure use of information and communication technologies in support of health and health-related fields, including health-care services, health surveillance, health literature and health education. eHealth uses modern information technologies in order to improve medical services. eHealth cloud is the implementation of cloud computing technologies in healthcare sector. It is the use of cloud technologies for the medical purposes. eHealth cloud can improve patient care, reduce operational cost and optimize IT resources utilization, it can be used to support researchers, national security and strategic planning; cloud allows to enhance security safeguards. In this thesis we proposed an e-health architecture within smart cities.
Breast sentinel lymph node cancer detection from mammography images based on quantum wavelet transform and atrous pyramid convolutional neural network
This research was conducted with the intention of locating the specific mass identification procedure that takes place after the noise has been reduced by identifying the salt and pepper, Gaussian, Poisson, and impact noises that are present in mammographic images. The purpose of the study was to identify these types of noise. As a consequence of this, it provides image morphology operators for accurate mass segmentation in mammographic pictures based on classification utilising Atrous Pyramid Convolutional Neural Network (APCNN) as a deep learning model, in addition to the noise-reduction strategy of Quantum Wavelet Transform Filtering. This is accomplished as a result of the aforementioned combination. This goal may be reached by using both of these methods in conjunction with one another. Using metrics for instance, maximal Signal-to-Noise Rate (PSNR) and Mean Squared Error ( MSE ) in lowering noise and detection precision, the QWT-APCNN hybrid technique is compared to conventional methods for the purpose of identifying mass areas. This comparison takes place with the intention of finding mass areas. These measurements are used in order to assess how well the hybrid approach was implemented. The technique that has been suggested has higher noise reduction and segmentation capabilities when compared to the most current breakthroughs in the area. For the goal of this study, it is expected to be able to empirically detect breast cancer, pinpoint the specific location of the masses, and categorise them as benign, malignant, or suspicious. This is intended to be accomplished to the greatest extent that is feasible. In this thesis, we implemented a novel way of deep learning using an APCNN that was built on a CNN. This approach is able to simultaneously extract features and classify data. A novel technique called APCNN has been developed that can simultaneously extract features and classify data. According to the findings, the approach that was suggested outperforms the alternatives in terms of a variety of metrics (such as accuracy, sensitivity, and specificity) at a rate that is 86.77%, 90%, and 90%, respectively. The fact that the proposed approach has these percentages demonstrates that this is indeed the case.
Doğru akım eğilimiyle ücretsiz hermit simetrisine dayalı spektral olarak verimli hibrit radyo frekansı ve görünür ışık iletişimi
To tackle the rapidly increasing demand for Internet services from mobile devices, network convergence is envisaged by integrating various technology domains. A recently proposed and promising strategy to indoor wireless communications is integrating the wireless fidelity (WiFi) and light fidelity (LiFi) namely a hybrid WiFi and LiFi networks (HWLNs). Such type of networks can handle the short comes of each standalone technology such as the spectrum congestion of WiFi and the limited area coverage of LiFi. In HWLNs, the WiFi sub-network follows the IEEE.802.11series standard applying quadrature amplitude modulation with orthogonal frequency division multiplexing QAM-OFDM to support high speed. While LiFi sub-network follows IEEE.802.11bb standard where OFDM-based modulation approaches are applied. In LiFi technology, light emitted diodes (LEDs) are applied as transmitters then only the signal intensity can be used to encode the information. Consequently, the signal needs to be real and unipolar at the input of LEDs. There are few spatialized approaches which account for this, including the most common PAM based discrete multitoned modulation (PAM-DMT) and direct current biased optical OFDM (DCO-OFDM). However, all of them achieve the real signal representation on the cost of halving the spectral efficiency as they impose the Hermitian symmetry. In this work, an optical OFDM approach named Hermitian symmetry free with direct current biasing (HSF-DC) is proposed to increase the spectral efficiency of LiFi subnetwork by 100%. The proposed HSF-DC scheme applies clipping, splitting, and reordering on the pulse amplitude modulation based OFDM (PAM-OFDM) instead of imposing Hermitian symmetry to achieve the real signal representation, then a suitable DC-biasing is added to ensure the positivity of the signal. The energy efficiency of HSF-DC is manifested in the fact that it has the least Peak-to-Average Power Ratio (PAPR) and the best Bit Error Rate (BER) compared with the counterparts. The proposed HSF-DC is then integrated with QAM-OFDM to build a spectrally efficient HWLN with gain of 1.3 over QAM-OFDM/PAM-DMT and QAM-OFDM/DCO. Finally, computer simulations show that the proposed hybrid QAM-OFDM/HSF-DC has the best BER performance over stand-alone QAM-OFDM, hybrid QAM-OFDM/DCO and QAM-OFDM/PAM-DMT.
Gelişmiş otsu eşiği ile çok düzeyli görüntü eşiği için aes crypto ile birleştirilen black wıdow optimizasyon algoritması
Thresholding images, a technique that uses histogram analysis, is an important aspect of the study of image processing. These methods an image's histogram and recommend optimal threshold settings for discriminating between image regions. Since thresholding achieves high accuracy and efficiency, it is widely employed in studies involving picture segmentation. Among the most significant thresholding techniques in image processing is called multi-level thresholding, and it incorporates the Otsu approach, which is utilized by a large number of people. While these approaches are effective and efficient, they are also very computationally intensive. Increases in the number of criteria employed cause these strategies to become less effective because of their increasing complexity and execution time. The Black Widow Spider Optimization Algorithm is one example of a metaheuristic algorithm that can be utilized as part of the Otsu threshold technique to help find good threshold values. These algorithms can determine, in a reasonable time, which thresholds are most effective for a given image. The suggested technique considers each threshold as a constituent or facet of a solution to the Black Widow Spider Optimization Algorithm in order to determine the optimal threshold value with minimal intricacy. Cryptographic techniques are gaining prominence as a means of guaranteeing safe data transit across a wide range of use cases. As more and more industries begin to rely on visual processes, protecting sensitive proprietary image data becomes more vital than ever. When it comes to protecting sensitive information, nothing beats the AES, a block cipher that has numerous benefits. In many image processing applications, multilevel thresholding image segmentation is popular. Thresholding uses gray levels to separate an object from its background in photo segmentation. Metaheuristic methods like the Black Widow Optimization (BWO) Algorithm can determine the Otsu threshold. The study's goal is to prove that employing an iris template may provide a cryptographic key that is both secure and difficult to compromise. An iris template or code is generated from the processed iris pictures and utilized for security purposes. During this phase, BWO method is used for image thresholding. For data protection purposes, we segment the iris from an eye image before encrypting and decrypting it using the advanced encryption standard (AES). Experimental results reveal that iris image encryption and decryption techniques were quick and secure. The results reveal that the suggested technique has a higher PSNR index value than the competing algorithms in 83.33 percent of the experiments, and a higher SSIM index value in 80 percent of the experiments. The suggested algorithm's thresholding performance on medical photos like brain tumours and human retinal images is demonstrated by an analysis of a set of pertussis images.
Magnetic resonance imaging (MRI) for brain tumor and seizures classification using recurrent neural network
This research work aims to utilize the developed and evaluated Magnetic Resonance Imaging (MRI) technique for the classification of brain tumor and seizures employing Recurrent Neural Network (RNN). The medical science in the image processing is an emergent area that has suggested many progressive methods in detecting as well as analyzing a specific disease. Brain tumors treatment is recently getting progressively more challenging owing to the intricate shape, structure and the texture of tumor. So, via progressing in the image processing, different methodologies have been suggested for identifying the tumors inside brain. The progression in such area made a need for searching more upon the methods and approaches evolved for the extraction of tumor. Therefore, an extraction system the tumor from the brain is suggested utilizing MRI images. Such method includes various procedures of image processing, like filtering, the removal of noise, segmentation, and morphological processes. Brain tumor extraction can be successfully achieved via conducting such processes upon MATLAB software. The cross-correlation is calculated between the changeable vector of a target and the zone of tumor for determining in what way the values of pixels of the zone of tumor are narrowly associated utilizing the image processing and the RNN method accomplishing 99.71% accuracy. The model being fit to the data till no additional remarkable reduction in the loss function value was performed upon the data of confirmation. It was trained and confirmed till (100) epochs upon precise (200) samples and (65) samples correspondingly with 10-cross fold validation. The measured classification accuracy of the brain tumor at the last epoch of the classifier training of validation data is virtuous.
Akıllı invertör kullanarak hibrit pv /rüzgar sistemlerinin derin sinir ağı denetleyicisi tabanlı güç kalitesi iyileştirmesi üzerine bir çalışma
Nowadays, the world is witnessing uncontrollable changes, such as global warming and climate change because of excessive fossil fuel utilization for transportation and power generation; therefore, the developed world is mainly concentrating on alternative resource development; for which, they have taken major research and development initiatives. Generally, certain alternative power generation sources, including wind, solar, and hydropower are not detrimental to nature. For this reason, solar and wind power have been declared as useful alternative energy resources, and besides, they are abundant. This thesis shows the investigation of the performances of wind energy systems and photovoltaic cells when weather conditions keep on changing. The findings of this research provide the basis for developing an advanced intelligent control system to maximize power generation. For renewable energy sources, the MPPT controller is essential because weather conditions are mostly unpredictable. This thesis has been written with a major objective to suggest a new algorithm, which is based on deep neural network (DNN), and to apply it for maximum power point tracking (MPPT). MATLAB was used to simulate this project for wind-based power generation systems and photovoltaic (PV) cells. An advanced DNN controller was developed for reducing the THD value and improving the output power quality of a microgrid-integrated hybrid wind/PV power generation system. The performance of the proposed system was tested and analyzed in several possible operating conditions using MATLAB to assure its functionality. In the end, we also analyzed the results of simulations using the IEEE 1547 standard.
5g mobil ağlarda büyük veri tekniklerinin uygulanması
With regard to the forthcoming requirements for mobile services in 5G networks, several new technologies have recently become the focus of leading-edge research. One of those is in-band full-duplex communications in processing data with big data based apache spark. The idea is to simply employ the same frequency band to simultaneously transmit and receive information, allowing more spectrally efficient communications when compared to the traditional half-duplex or out-of-band full-duplex counterparts. By breaking a long-held assumption in wireless communications, in-band full-duplex may double the throughput or reduce by half the allocated bandwidth in future transmissions. Nevertheless, the self-inflected interference, that naturally arises, poses the major problem to the use of this technique, enhanced when in presence of system impairments with apache spark. Therefore, this thesis aims to study wireless communication in 5G network with use of big data based apache spark. Particularly, apache spark suppression filters are studied, as well as feedback adaptive filtering is proposed for relay systems with multiple-input multiple-out (MIMO) antennas. Furthermore, the orthogonal properties of large-scale arrays are resorted to, such that an extra level of mitigation is achieved. In this case, the relay system energy efficiency is maximized by finding the optimal transmit powers, while maintaining a certain individual link quality. For this scenario, the effect of massive MIMO is likewise addressed, and an algorithm that maximizes the system total achievable rate is derived. The network of wireless communication requires low latency and high availability of 5G mobile network connections to enable co-operation between the networks and infrastructure. The performance and functional properties of the network slices can be modified optimally with network slicing, to meet these requirements with apache spark. These new 5G technologies, however, have pressing security challenges to secure data integrity and privacy in critical wireless communications requiring extensive research before implementing these technologies into use. This research examines the new possibilities 5G networks offer for wireless network communication with high bandwidth of 90 GHz. The prediction of 5G network bandwidth maximum recoded at 90 GHz on 5 nodes of apache spark at 80% of data trained with total 12 nodes. The main question to answer is will the 5G technology offers sufficient performance for wireless communication. Keywords: 5G, in-band full-duplex, MIMO, big data, wireless communications, apache spark, spectrum allocation.
Classification of normal and abnormal activity in SCADA system using multi-kernel SVM
SCADA sistemleri veya İngilizce kısaltması için denetim, kontrol ve veri toplama sistemleri, basit terimlerle her türlü endüstriyel işlemi uzaktan kontrol etmek ve izlemek için oluşturulan kontrolör ve bilgisayar ağları olarak tanımlanabilir. SCADA üzerindeki normal aktivitenin özelliklerini ve bir sınıflandırma sistemini beslemek için anormal aktiviteyi çıkaran ekstraksiyon yöntemi, özellik çıkarma aşamasının sağladığı verilere göre aktiviteyi normal ve anormal olarak sınıflandıran sınıflandırma şeması.
Sinir ağlar kullanarak örnek tanıma
Abstract— Due to its various applications, such as security systems, medical systems, entertainment, etc., face recognition has also been identified as one of the main research topics. The preferred method of human identification is face recognition: natural, robust and non-intrusive. A wide range of systems require reliable personal identification schemes to either confirm or determine the identity of a requester. The purpose of these schemes is to ensure that only a legitimate user and no one else accesses the rendered services. For example, secure access to buildings, computer systems, laptops, mobile phone and ATMs is included. These systems are vulnerable to an impostor's will in the absence of robust personal recognition systems. This article has developed and shown the human face identification system using artificial neural networks, which reflects that the face recognition rate for 40 individuals shows results for 400 frames in the AT&T database at 85.5 percent.
Enhanced secure communication schemes for machine-to-machine and vehicular ad hoc networks
In this thesis, we examined several Machine-to-Machine (M2M) and Vehicular Ad Hoc Networks (VANETs) authentication and revocation schemes. According to our analyses, we introduce new schemes without flaws of examined designs. A password-based authentication for mutual authentication is suggested in the first analyzed M2M communication system. Furthermore, their suggested secure channel setup protocol utilizes symmetrical encryption and single way hash algorithms and considers that portable consumer devices or smart home networks are using their secure channel architecture. We propose that the missing part of the current M2M secure communication system be completed. An improvement of the system can be made by protecting privacy and altering the messages. Elliptic Curve Diffie Hellman (ECDH) cryptography based secure key sharing scheme, introduced in both initial setup and key-injection stage to provide safe client enrollment, device key change, and home gateway network's connection stages. The second studied VANET secure authentication and revocation system partly replaces Certificate Revocation Lists (CRLs) monitoring technique with the Hash-based Message Authentication Code (H-MAC) cryptogram verification. Public Key Infrastructure (PKI) and CRLs used communication systems have critical delays during the monitoring of CRLs. This latency has an essential impact on VANETs taken by PKI. We evaluated the VANETs system and resolved common open problems. We also conducted missing components of comparable systems with efficiency improvements. These can occur through revocation key sender verification and revocation version validation to protect sensitive assets from invalid updates, address privacy preservation with keyed trimmed H-MAC based pseudo ID creation, set message identity to perform sensitive information transmission, lastly extract, and integrate applications to execute high-speed revocation. Furthermore, our reforms prepared for system reliability, durability, and MITM, substitution, false message, message modification, user manipulation, and successive response attacks resistance with anomaly detection. For M2M scheme, we simulated both proposed and analyzed plans for performance, network congestion, and resource usage. Also for VANETs scheme, we simulated three cases the standard, the proposed and analyzed models. We analyzed schemes for network congestion and performance. According to our simulation results, we proposed efficient schemes for both M2M and VANETS.
LTE OFDMA ağları için deneyim tabanlı dinamik hücreler arası bantgenişliği paylaşımı
In recent years, developments in telecommunications systems have led to the appearance of new demands. For wireless communication systems users, data transmission speed and data sharing have become the most important issues. Thanks to the technological advancements that have been experienced every day, the data transmission speeds have reached unpredictable dimensions. When considering the users' demands and limitation of the frequency spectrum, it is clearly understood the importance of effective way of resource allocation, resources distribution and efficient use of existing resources. At the same time, scientists want to avoid intra-cell interference. For this reason, orthogonal frequency division multiple access (OFDMA) in Long Term Evolution (LTE) can minimize intra-cell interference using sub-carriers that are orthogonal to each other. But, inter-cell interference, limits the downlink performance of cellular systems. To minimize inter-cell interference, several interference cancellation techniques have been analyzed and compared. One of these techniques has been analyzed in detailed and some improvements made in this technique. The first objective of this technique is to minimize inter-cell interference by developing the frequency reuse technique. Also, it aims to increase the throughput and signal to interference plus noise ratio (SINR) of the entire setup and prevents the overload of the cells by allocating and reallocating the bandwidth dynamically between the cells. The last target of the proposed technique is to progress a fairness scheduler algorithm among users by optimizing the base station schedulers that is called experience-based packet scheduler (EBPS). Finally, combining all these targets, I proposed the algorithm called Experiment Based Dynamic Soft Frequency Reuse (EBDSFR) in this thesis. First of all, EBDSFR technique has been compared with Reuse-1, Reuse-3, Fractional Frequency Reuse (FFR), and Soft Frequency Reuse (SFR). Secondly, I compared our proposed technique with the Dynamic Inter-cellular Bandwidth Fair Sharing FFR (FFRDIBFS) and Dynamic Inter-cellular Bandwidth Fair Sharing Reuse-3 (Reuse3DIBFS).
Evaluation of water state using combined temporal remote sensing and image processing techniques
Remote sensing (RS) has significant role for detecting and monitoring different objects on the earth through the analysis of data which acquired by various types of sensors. Monitoring capacity and level of water resources is an important matter for different fields. This work demonstrated the efficiency of MATLAB software to monitor the water state in Mosul Dam Reservoir which is located in the North West of Mosul city over the period of years between (1986-2017). In this work, various digital image processing (DIP) algorithms were performed and Multi Spectral (MS) temporal Landsat data were acquired over time period. Two main steps were processed, preprocessing which illustrated how to prepare the Landsat's data to be more accurate for extracting the information, and post processing step implement various digital image processing techniques that included integration and hybridization between two processing methods in the stage that led to gain the target of this work with high performance. Different objective functions criteria's were performed to test the performance and the accuracy of the proposed work, regarded to the adopted method. In other word, Peak Signal To Noise Ratio (PSNR) and Mean Square Error (MSE) were utilized for the obtaining the performance of coastline detection as s first stage. Furthermore, the Assessment Accuracy (AC) depending on different values had demonstrated the performance of hybrid classification method for classification the quality of Water Lake. Consequently, it is infer that this work gave a futuristic view for the status of water state monitoring through combined applications related to temporal data and image processing. As well as, its concentrated on the evaluation of water state areas in Mosul city and Mosul Dam Lake due to its importance for providing the electricity to the Mosul city. It is also considered one of the most important tourist lakes in Iraq. So the attention about this area is significant for most researchers who interested on water resources studying.
The impact of social media on university students in Iraq and Turkey
Today we live in the technological revolution that began to dominate the most details of our daily lives and became the important concern the most widespread factor is the means of social communication consequently, in this study, we decided to shed light on the general effects, academic, political and business by measure difference impacts on the students of universities in Iraq & Turkey, and I have chosen three Iraqi universities (Kerbela University in the Middle Euphrates, UOITC University in Baghdad and Tikrit University in western Iraq) and three Turkish universities (Altinbas University, Marmara University and Karabuk University) where we put it, in to consideration, cultural and social differences accordingly, Data was collected from survey distributed over (34) questions. We tried to cover most of the common student's who uses social media and the impact on them. The questionnaire was distributed through (100) survey paper to each Iraqi university. The total number of participants (201) in the Iraqi universities distributed to (77) participants from the University of Kerbala, (50) participants from the University of Tikrit and (74) participants from the University of Information and Communication Technology (UOITC) and for Turkish Universities we published an online survey but the participation was somewhat weak. So we used a paper survey to increase the number of participants in the three Turkish universities. The total participation was (50) participants from Altinbas University, (38) from Marmara University and (33) from Karabuk University.
Mobil jet kasa tasarımı
Mobil ödeme sistemleri her geçen gün daha gelişmekte ve kullanıcıların işlerini daha kolaylaştırmaktadır. Bu çalışmada var olan ödeme sistemleri ele alınıp, avantaj ve dezavantajları ilişkilendirilip daha efektif ve kullanışlı yeni bir mobil ödeme sistemi geliştirilmiştir. Bu yeni sistem büyük market zincirlerindeki ödeme sistemlerini hedef almaktadır. Tüketicilerin günlük alışveriş işlemlerini kolaylaştırıp, daha hızlı, daha güvenli ve daha rahat gerçekleştirmesini sağlamaktadır.
Mobil jet ödeme sistemi yazılım altyapı oluşturulması
Ödeme sistemleri her geçen gün çeşitlilik kazanarak yaygınlaşmaktadır. Her ödeme sisteminin temel amacı müşteriye daha iyi bir alışveriş deneyimi kazandırmaktadır. Bu sistemde de amacımız sadece müşteriye daha hızlı, güvenli ve kolay bir alışveriş deneyimi sağlamak değil ayrıca bu sistemi kullanacak olan mağaza / marketlerin de faydalanabileceği satış öncesi ve sonrası faydalı ve kullanışlı özellikler sağlamaktadır.
Yapay sinir ağının benzer analizlerin sınıflandırılmasında farklı tekniklerin karşılaştırmalı analiziyle kullanılması
Sentiment Analysis means identifying the favorable, negative or neutral opinion or reviewer opinions expressed in a piece of job. In social media surveillance sentiment assessment is helpful to automatically characterize the general impression or mood of the customers as replicated on their social media for a particular brand or business and determine if they are regarded favorably or negatively on the Internet. This article examines the machine-based learning approaches to feeling assessment and highlights the key characteristics of methods. Prominently used techniques and methods are Naïve Bayes, Maximum Entropy and Support Vector Machine, the most near-neighbor classification of Machine Learning-based feeling assessment. Naïve Bayes ' depiction is quite easy but does not give rise to wealthy assumptions. The hypothesis that characteristics are independent is too restrictive. Maximum Entropy estimates the distribution of probability by information, but it does well only with dependent characteristics. For SVM the kernel is correct, but the way to deal with multi-class issues is not standardized. A method which combines neural networks and fuzzy logic often is used to improve the efficiency of correlations and dependencies between variables.
Büyük veri analizi via bulut bilişim
The primary importance of the data lies in the fact that it provides information to take correct decisions. The data becomes unhelpful if we are unable to extract information out of it. The process of analyzing data means the ability to extract information from a specific data set that a user requires to make the decision-making process easier, and besides, it is helpful to make accurate decisions. There are many obstacles in Relational Database Management Systems (RDBMS) because of their old design. It is very difficult to meet the ever-changing and continuously increasing needs of the information technology systems. For example, these systems find it difficult to handle data in exabytes or zettabytes. Such large and unstructured data is termed as "Big Data." This thesis explains how to process magnanimous and difficult-to-handle unstructured data in order to control and analyze it. This problem has been highlighted when different solutions were introduced, which include the emergence of Cloud Computing (CC) that provides a useful alternative to the traditional systems, also known as massive Data Analytics. A large number of students, engineers, and staff members of Altinbas University participated in a survey as a part of this research. Some Libyan participants also participated in the survey. The participants were asked to fill in a questionnaire that had 30 questions focused on how Cloud Computing services are helpful to handle and analyze big data. We found almost of them participants had no knowledge of cloud computing services; so, it became a challenge; however, this thesis throws new light on the subjects like big data and cloud computing. A big data analysis project shows how to analyze one of the largest datasets in a quick and easy way. By the end of the thesis, we aim to provide a complete definition with a clear example that encourages users to use Cloud Computing Platforms instead of traditional memory. Keywords: AWS, Big data, Cloud Computing, GCP, RDBMS, MS Azure.
A novel service-oriented architecture for implementing the smart cities in terms of complexity of smart services
The considerable increase of the urban population around the world has been a great challenging to the governments and the existing management of the cities with regards to the high demand of the public services, urban road network, more parking, health services, education, electricity, environment etc. The modern concept of the smart cities is strongly believed as one of the innovation strategies to solve these major issues in the most efficiency and cost-effective way by delivering the smart services faster, more reliable and safer to allow sustainable development and provide the smart living, smart economy and smart environment to the smart citizens. In this thesis , we will provide the general concept stressing on what the definition of the smart cities actually is. Finally, we will describe in the fully detail of the three S-Dimensions used to classify whether the city considered as the smart city or not. Keywords: Smart cities, S-Dimensions, ICT.
Comparison the reliability of visible light communication usin whit leds based MIMO with RGBY LEDs based wdm
Visible light communications VLC for indoor application nowadays rapidly increased especially when LEDs be used for illumination, that give the lighting system another task which is high data rate communication, and some time positioning task. In this thesis two typed of VLC systems simulated depending on transmitters type (white LEDs and RGBY LEDs). Modulation techniques effect also studied. CMO, DCO and GLID modulation schemes simulated. Rotating angle of effect studied, results recorded and discussed. Conclusions illustrated and future works also suggested. İç mekan uygulamaları için görünür ışık iletişimi VLC günümüzde özellikle aydınlatma için LED'ler kullanıldığında, aydınlatma sistemine yüksek veri hızı iletişimi olan başka bir görev ve bazı zaman konumlandırma görevi veren hızlı bir şekilde artmıştır. Bu tezde, vericilerin tipine (beyaz LED'ler ve RGBY LED'ler) bağlı olarak simüle edilen iki tip VLC sistemi. Modülasyon tekniklerinin etkisi de incelenmiştir. CMO, DCO ve GLID modülasyon şemaları simüle edilmiştir. Dönen etki açısı incelendi, sonuçlar kaydedildi ve tartışıldı. Sonuçlar ve gelecek çalışmalar da önerilmiştir.
Hücresel ağlarda orantili adil zamanlamanin ikinci mertebeden ergodik analizi
This dissertation discusses a method based on Gaussian Based approximation to evaluate the effctiveness of the Propor tional Fair Scheduling in cellular network wireless networks. The Propor tional Fair ness Scheduling is found ver y useful in the 4th Generation 4G and the 5th generation 5G as it balances the effiency of the throughput and the fair ness of the users. Never theless when channel conditions become non-unifor m the users tend to be more active, thus their scheduling problems compounding especially in crowded and fast changing environment. In this regard, by assuming that the users' rate distr ibutions are Gaussian, a model is developed which facilitates easy calculation of mean throughput and fair ness measures. In this disser tation, the Gaussian-based analytical approximation and the system-level simulation studies are investigated independently, and the simulation platfor m implemented in MATLAB is subsequently used for validation and comparative evaluation of the analytical model. Simulations take into account channel quality indicator (CQI) provision, Rayleigh fading, movement of users, and a pr ior ity based modifiation of the scheduler. A par ticular simulation studied three users: the same, a propor tionately changing and a completely diffr ing users. Parameters such as average user and system overall throughputs as well as jain's Fair ness and CQI load profies were evaluated for each case. Moreover, enhanced PF scheduler model with weights used for user load balancing was created and evaluated. The analysis of the simulations shows that although such adjustment could help improve individual parameters this could lead to an inevitable even distr ibution of resources between users. Simulation Models show g reat per for mance, with PF method, when the models based on Gaussian approximation are used as they tend to enable high reliable per for mance, and with ease, minimized the resources needed for scheduling strategy assessment. The results of this study will contr ibute to the improvement of such systems in 5G and 6G where scheduling and resource allocation mechanisms should be fur ther developed.