Theses supervised by Yrd. Doç. Dr. Sefer Kurnaz
18 theses · Altınbaş University
Proxy internet hizmetleri ve nesnelerin interneti için makine öğrenimi
The last decades have shown a rapidly growing number of connected devices of the Internet of Things (IoT). Accordingly, more significant access is needed to the internet service to operate the devices. However, access to the services became exhaustive on IoT devices due to limited available resources. This limitation motivated proposing of lightweight protocols, (for example, constrained Application Protocol (CoAP) and MQ Telemetry Transport (MQTT). This research uses these lightweight protocols and proposes a novel system capable of proxying internet services by using their protocols and requirements for IoT devices. On IoT devices, the experiments demonstrate a significant decrease in resources consumption in terms of network bandwidth, time, memory, and energy. Consequently, the efficiency of the devices is significantly improved. Furthermore, the same (futher) tasks can be performed using the less (same) resources devices.
FPGA tabanlı PLC tasarımı ve kullanımı uzaktan kumandalı indüksiyon motor sürmek
As a result of continuous development, the importance of variable speed drive has risen dramatically in recent years. Over time, the demand for modular and more standardized drives has increased. Thyristor technology, high-speed digital switching systems, microprocessors, Programmable Logic Controller (PLC), microcomputers, and microcontrollers have all contributed to the popularity of drive technology. High-performance and reliable drives are already being used with advanced technology. Drives frameworks with low power utilization and little size are popular today to satisfy the needs of high voltage, minimal expense execution, and remote observing through information procurement frameworks. A significant initial phase in creating complex plan and control techniques for remote-found drives has been made with the present PLC framework. Enlistment engine boundaries will be digitalized with high exactness and speed utilizing the proposed framework's re-configurable programmable gadgets and Simple to Computerized Converters with sufficient transformation speed. FPGA will be utilized for LCD connect with the FPGA to show different engine boundaries utilizing the simultaneously working model that will be assembled. When it comes to implementing the VHDL-coded hardware modules, the FPGA will be employed by Xilinx.
Derin öğrenmeyi kullanarak zararlı yazılım sınıflandırması
In this study, the performance of different machine learning algorithms was evaluated for detecting malware based on the permissions used by Android applications. The study aimed to compare the performance of four different algorithms: k-nearest neighbors (KNN), a convolutional neural network (CNN), a convolutional neural network with long short-term memory (CNN-LSTM), and a convolutional neural network with the gated recurrent unit (CNN-GRU) by using the Python programming language. The dataset used in the study consisted of binary vectors representing the permissions used by each application and was labeled as either malware or non-malware (benign). The performance of the algorithms was evaluated using standard metrics such as ACC, PREC, and recall. The results showed that the (CNN-GRU) algorithm had the highest accuracy with 95%, followed by the (CNN) algorithm with 91%. The (KNN) algorithm had an accuracy of 87% while the (CNN-LSTM) algorithm had an accuracy of 90%. All four algorithms had similar precision and recall scores, with the (CNN-GRU) algorithm having the highest F1-score of 94%. The loss values for all four algorithms were also calculated and compared. In conclusion, the results indicate that the (CNN-GRU) algorithm performed best in terms of accuracy and F1-score, followed by the (CNN) algorithm. These findings can help inform future research in the field of malware classification using deep learning.
Yenilikçi dl ilkeleriyle güneş pv sistemlerinden üretilen elektrik gücünün tahmini
This master's research is implemented to examine and explore the significant contributions and substantial relevances of novel DL algorithms in making predictions with considerable effectiveness, accuracy, reliability, and trustworthiness. The work addressed a case study comprising two solar PV systems in India. Datasets were defined, and two performance evaluation approaches were used to examine the accuracy of the three DL models: LSTM, CNN, and LSTM-CNN. The numerical simulations and mathematical analysis in this work revealed that the LSTM, CNN, and hybrid LSTM-CNN models offered higher accuracy in forecasting clean electrical power produced from the two solar PV systems. Nonetheless, the accuracy of the hybrid scheme provided the most significant accuracy rates compared with the first two algorithms. The research findings also indicated that the values of MAE linked to the training of the three algorithms were more considerable than the MAE amounts related to the testing process across all the epoch ranges. Besides, it was found that the MAE connected with training for the three algorithms started at a maximum value. Then, it declines until it reaches a lower value steadily for a more extended epoch range. However, the MAE value after declination was still larger than the MAE rates of the testing process. Further, the numerical outputs confirmed that the RMSE value of the training and testing procedures for the three algorithms had similar behavior to the MAE in forecasting the clean electrical power of the two solar PV systems.
Yenilenebilir enerji tabanlı akıllı elektrik şebekesinde yapay zeka kullanarak hata tahmini
The concept of intelligent energy networks (Intelligent Energy Networks), or Smart Grid (SG) has been developed aiming to make the EPS operation more robust, economical, reliable, available and automated. New technologies, applications and smart devices have been developed, mainly to make the distribution layer smarter. However, an EPS must have an integrated intelligent operation, aggregating all functionalities at all levels of the EPS with the aim of improving the EPS's performance in terms of economics, quality of service and technology. The use of these resources makes it possible to make the operation of the electrical system more flexible in the contexts of generation, transmission and distribution, reaching new levels of quality for these services. One of the main impacts that EPS have on current systems is in the context of energy generation The objectives of this work are to analyze and develop methods for predicting energy demand and supply in an SG in the context of Smart Grids.
Akıllı tanımlama sistemi tabanlıiris deseni tanıma üzerine
The identification process is one of the important processes to achieve safety, which is used daily in many areas such as the attendance system, various Internet or social sites, personal accounts inside banks and other important areas. In this article, the model is a genuine model mode as a marker of the character of the individual suggested on the SIFT technique. To tackle identity theft, poor picture quality, noise and volatility of the ambient world, we have defined the iris patterns and features for each person. The Hamming distance algorithm was used on a dataset of 50 individuals to figure out how close new entrants to the dataset ought to be to a current person's entry. 98% of the value, the bogus assent rate, and the refusal pace of 0.15% and 0.25%, separately, yielded the aftereffects of the application strategy.
C-BAND İÇİN ÜÇ BANTLI ÇİFT H-SLOT MPA TASARIM
In several significant wireless implementations, the MSP antennas. Double H-slot patch 3-band antenna in C-Band is implemented in this article. The radiation productivity, including radiation patterns, antenna benefits, VSWR and rerouting losses, was simulated and described using ANSOFT HFSS Ver. 17.2. The design of this antenna is easier than other 3-band antennas and only consists of a substratum, a soil and a coaxial feeder string. I handle in C-band (4-8 GHz) and we have 3 frequencies in the sector (4.01 GHz, 5.81 GHz, and 6.55 GHz). On the Rogers RT/duroid 5880/ ℇr 2.2, the spreading patch is tumbled and the Hight is 15748mm for the substratum pattern. The fifty-ohm antenna is fed with coaxial feeding technology. The lack of return of the Antenna improved to 27.13 dB (4.01 GHz) and 5.81 GHz (-21.78 dB), while 7.754 dB (6.55 GHz) and 4.01 GHz, 8.357dB (5.81 GHz) and 7.232 dB (6.55 GHz) suggested the return loss for that Antenna. Simultaneously. The simulation computer displays VSWR 3.95-4.05 GHz, 5.75-5.88 GHz, and 6.5-6.61 GHz.
Sağlıkta blockchaın: Karma diş dönemindeki çocuklar için diş sağlığını iyileştirmeye yönelik akıllı sözleşmeler
Privacy preserving is a matter of great importance in many sensitive fields including financial and healthcare areas. As the blockchain technology has achieved an obvious success in privacy protection at the financial level when the Bitcoin has been introduced in 2008, there are considerable attempts and later on successful projects to introduce the blockchain technology to the healthcare fields, where the patient privacy and access control to the personal information is at high priority. The aim of our thesis mainly considers the introduction of the smart contracts that essentially depend on the technology of blockchain to act as a method for preserving patients' privacies along with improving the reality of dental healthcare providing services. The main focus will be on the treatment plan of the patient by building a smart contract system between multiple partners. Firstly, the problem will be presented that is why there is a need for implementing blockchain system of smart contracts to improve dental healthcare providing? And why we need to provide digital organizing of treatment plan? Secondly, The Blockchain is discussed as a relatively new technology. Thirdly, the design to solve the problem will be proposed and the system of smart contracts will be built, with explanation of the way by which the system will meet the requirement to give the solutions. Finally, there will be a discussion of how proposed design will be implemented. By using "the five types models" our proposed smart contract system is built combining the advantages of the blockchain technology with the smart contracts benefits to provide decentralized, tamper resistant, and privacy preserving properties. The programming language that we have used is Solidity along with proof of authority algorithm on Ethereum blockchain. The proposed blockchain-based smart contract system can provide high level of privacy preservation in the field of patient records in addition to the digitalization of healthcare data.
Bulanık mantık kullanılarak dağıtılmış kontrol sistemlerinin yönetimi
The objective of this project is to implement an automatic system in one of the stations of a real production line of oil, in the building of the Higher Technical Oil refineries. Likewise, the objective is also to implement a SCADA type control system to control and monitor the operating parameters of the station using fuzzy logic, and to carry out a statistical control of the quality of the parts processed by the system, showing a series of statistical tools in time. real that allow to verify the correct operation of the process within the established quality parameters.The station to be automated is Station 1,and the SCADA system will work on a Computer connected to the same network.
Konum koruması için karşılıklı politika kablosuz ağlar
To maintain area protection for the Wireless Sensor Network (WSN) control frameworks, use K-type anonymisation to work together to alert a worker about unknown common regions. This dark location k is a location that is, nonetheless, covered by k persons. In any event, we have defined an attack model to demonstrate that settled grouped targets are still a safety danger, on the assumption that the opponent may collect certain areas covered by people under k without taking account of nameless protection requirements. The WSN Mutual Privacy Convention (REAL) has been proposed in this article. You actually require a sensor hub to adjust your sensor districts separately into a range of total areas that do not cover and are very accurate. In order to deal with three real issues, which are autonomous guidance, regular resources and high precision, we also have cultivated a State moving structure, lockout system and time-replacement. REAL yield and contemporary practices using theoretical testing have been compared. Results reveal that REAL keeps the sites concealed, answers inquiries more accurately and reduces the gadgets and expenses of accessibility
İç mekan uygulamaları için mikro şerit yama anteninin performans artırması
There are many structures of the antennas utilised in wireless communication applications, one of the most widely known and utilised is the Microstrip Patch Antenna (𝑀𝑃𝐴). This antenna is most recognised due to its benefits in terms of geometry configuration that makes it fitting for various of the communication via distance application. The small weight of this antenna structure and the appropriateness for mixture with many of the microwave applications is very useful as well as required. This antenna has flawed by the narrow bandwidth and the low or the poor gain issues. Consequently, to utilise this antenna properly the mentioned issues should be optimised to boost the antenna performance. In this study, we will introduce a design of an MPA with a rectangular patch shape at an operating frequency of 60GHz employing the finite integration technique that is offered by the Computer Simulation Technology (CST) antenna simulation package. This frequency band is chosen due to the free license operation and the ability to get a higher data rate for the indoor application. In order to get the best connection and avoid mismatching issues, the inset-fed scheme is used to feed the simulated antenna; this is the method easy to manufacture and provides naivety modeling and impedance matching. In order to optimise the simulated antenna realisation, the mushroom-like Electromagnetic Bandgap (𝐸𝐵𝐺) square cells are introduced. These cells are implanted over the antenna substrate and connected with the antenna ground plane through the vias. Finally, a performance comparison between utilising the mushroom-like 𝐸𝐵𝐺 structure and the Frequency Selective Surface (𝐹𝑆𝑆) will be studied.
Yazılım tanımlı ağ ile kablosuz sensör ağı tasarlayın
This paper present design to wireless sensor networks based on Software Define networks .in this paper we provide some examples that show how architecture can use for networking .We also identify how to design WSN network with SDN controller and how to evaluate the performance of this network. Our design of WSN network consist of one sink node and twentyfour WSN node. The network is first tested without using SDN controller. Then the WSN network is operated with SDN controller. Both scenarios are evaluated by studding and analysis the main performance parameter (Received packet, received power, Throughput) and a comparisons is made. The received packet of WSN with SDN controller is increased by (53%) compared by received packet of WSN network without SDN controller. The throughput of WSN network with SDN controller is increased by (53.2%) compared by throughput of WSN network with SDN controller. The received power of WSN network with SDN controller is increased by (3%) compared by received power of WSN network without SDN. Also in this paper we implement networks of WSN and SDN using Real Node (Z1 mote) . From the last network we can see that the we can see small different in result between Real Network using (Z1 mote ) and simulated node Throughput is increased due to the increase in receive packet
Denir inanç ağını kullanarak öğrencilerin performans tahmini
Education research involving techniques of data mining is growing rapidly. Data mining techniques are known as educational information mining to explore hidden knowledge and patterns of student performance in educational backgrounds. This work seeks to develop an academic prediction model for the student for the UCI (University of California Irvine) machine learning repository dataset selected in this work for prediction. Internal marks are a combination of attendance marks, average marks from two examinations and marks of assignment. The teachers can therefore classify students and start predicting their performance at an early stage. To improve performance over time, systemic approaches can be adopted. Better results can be expected at the final examinations due to early predictions and solutions.
Gelişmiş bilgisayar ağında cisco packet uygulaması
In the process learning computer network systems, the use of virtual laboratories is very important. In this study, we also implemented Cisco Packet Tracer, which enables us to work on test scenarios without using any physical components virtually to design an advanced computer network. The Cisco Packet Tracer is used not only to simulate computer networks but also to learn computer networks. This study consists of fifth parts. In the first chapter, we briefly introduced the Cisco Packet Tracer, the aim, the importance, the assumptions, and the limitations of the study is explained. In the second chapter, conceptual explanations of the components necessary for designing a computer network are made. In this part, the configuration of the computer network, the types of topologies, network technologies and architecture, network cable systems and their components are described. In the third chapter, a program system was created for an exam that measures the capacity and potential of the participants about the computer network and their information about the Cisco Packet Tracer. Also, the map and rules of the computer network structure were determined to design a computer network in a 6-storey hotel. In the fourth chapter, in the framework of the specified network rules and map, the configurations have been made completely and the advanced computer network has been established using the Cisco Packet tracer. Finally, in the fifth chapter, five scenarios were implemented on the network in order to prove the network in the hotel is working well. The scenarios were about transfering packet between two floors, using the command line interface(CLI) between any two devices in the network, implementation of the hotel's website, making a telephone call between phones located on several floors of the hotel and using surveillance cameras to monitor all the hotel floors and connect these cameras to the network. The implementation of all these scenarios were showed that, the simulation network system was operated without any problem
TPS:Irak okulları için gelişen web tabanlı ev ödevi izleme sistemi (öğretmen,ebeveyin,öğrenci)
The world wide web affects education and our live in many ways. Nowadays, web-based homework has become widespread in several countries. Web-based homework systems in developing information technologies have many advantages compared to paper-pencil homework. The use of web-based homework systems eases the work of teachers, provides comfort to students, and ensures that parents are aware of their children's situation. The aim of this study is to create a new web based homework system (TPS teacher parent's student system) produce for Iraq schools. This system will be divided into three sections.The first section will be for the teachers. Teacher will be able to create homework, upload any video to illustrate their lessons, receive questions from students through the chat about the homework and the teachers can see a report of students performance. The second section is for students: the students can answer questions through the interface and can see the result immediately. The third section will be for parents, parents can follow their children by receiving reports about their children performance. In the first chapter of this study, the importance and the aim of the study were mentioned and the related researches and applications made in this area were given. In the second chapter, the web development tools such as JavaScript libraries, programming languages and databases required for the system to be developed and the Xammp server to be used in this study as a local server have been introduced. In the third chapter, the details of a new web based homework system (TPS teacher parent's student system) developed for Iraq schools are explained. Finally, in chapter fourth, presented comparison the TPS system with another web based homework system. Keywords: TPS Teacher,Parent,Student System e-Homework
En iyi uygun bulut bilişim hizmeti sağlayıcısı seçmek için çok kriterli karar destek yöntemleri arasında karşılaştırma çalışması
Cloud Computing, storage, processors, memory, etc. It is a special form of distributed computing, where resources are completely abstracted from the consumer. The number of Cloud Service Providers (CSPs) that offer the computer as a service has recently increased and many customers need to interact with unknown service providers to perform transactions. In such open and confidential environments, trust helps to improve consumer confidence and provides a reliable environment for them [8]. A trust-based ranking system can help them choose between services according to their needs. In this research, multi-criteria decision making methods (AHP, ANP, TOPSIS, ELECTRE, PROMETHEE and VIKOR) were used make rating for the providers according to their infrastructure parameters. Then a comparison algorithm invented to compare the methods ranking results. According to our case and used methods "Rackspace" provider was the most agreeable one between the others.
Verı madencılığı teknikleri kullanarak kalp hastahğı tanısı
Lately, large masses of data have been generated due to the ongoing approaches in biotechnology and fitness sciences areas. It combines clinical information and genetic data which included in Electronic Health Records (EHRs). On the other side, it is required to recognize symptoms, which can wrongly convince the human health in addition to placing economic burdens on their shoulders, in an early stage to avoid many difficulties. Lately, several data mining procedures have played a vital role in developing automated operations that can identify syndromes efficiently and correctly. In this thesis, we satisfy some of the research disciplines that have employed either the data mining procedures for identifying symptoms. Additionally, a set of well-known data mining methods including Decision Trees (j48), Naïve Bayes, Multilayer Perceptron (MLP), and Random Forest (RF) has been assessed in performing the classification task using a publicly available heart diseases dataset.
Veri madenciliği tekniklerini kullanarak ilaç etkileşiminin düzenlenmesi ve sınıflandırılması
Nowadays, there are thousands of approved drugs that can be used for treating people who have medical problems. Therefore, drug Warnings and Precautions section is intended to identify and describe a discrete set of adverse reactions and other potential safety hazards that are serious or are otherwise clinically significant because they have implications for prescribing decisions or for patient management. In this thesis, proposed tool takes the patient personal information, such as gender, age, pregnancy, the current diseases that the patient suffers from, the current drugs, and the newly diagnosed disease as input and returns the suggested drugs that suit the current patient state. The proposed tool consists of two main phases: data acquisition and data processing. In data acquisition phase, drug information has been collected from different sources including DINTO, DrugBank, RxNorm, and FDA. Moreover, a number of test cases have been designed for evaluating the performance of the proposed tool in different scenarios. The test cases have shown the efficiency with accuracy 96% and effectiveness of the proposed tool