114
Archived Theses
0
DOIs Assigned
0%
DOI Rate
Discipline
FPGA-tabanlı, steryo görmeye sahip bir robotta üç boyutta eş zamanlı konumlama ve haritalama
Konum ve harita bilgisi mobil bir robotun başarımında etkili olan önemli faktörlerdendir. Bu çalışmada, stereo kameraya sahip mobil bir robotun gerçek zamanda ve eş zamanlı olarak konumlama ve haritalama yapmasını sağlayacak genişletilmiş kalman filtresi için FPGA üzerinde paralel bir mimari oluşturulmaya çalışılmıştır. Sistemde, görüntülerde bulunan köşeler özellik noktaları olarak kabul edilmiştir. Stereo kameralardan alınan görüntülerdeki köşeler eşleştirme ve 3 boyutta konum tespiti aşamalarından geçirilerek köşelerin 3 boyutta robota göre konumu hesaplanmaktadır. Köşelerin 3 boyuttaki konumları genişletilmiş kalman filtresine ölçüm güncellemesinde kullanılmak üzere aktarılmaktadır. Robotta bulunan eyleyicilerdeki konum değişimi ise filtrenin tahmin aşamasında kullanılmaktadır.Çalışma Xilinx'e ait Virtex-5 XC5VLX50 çekirdeği üzerinde gerçekleştirilmiştir. Görüntüleri algılamak için Omni Vision'ın OV7720 dijital kameralar kullanılmıştır. Görüntüleri depolamak için Micron' a ait MT4HTF3264HY-53E DDR2 SODIMM bellek kullanılmıştır.
Channel adaptive user cooperation strategies for fading wireless channels
ABSTRACTCHANNEL ADAPTIVE USER COOPERATION STRATEGIESFOR FADING WIRELESS CHANNELSEDEMEN, ÇAĞATAYIn this dissertation, a channel adaptive method is proposed for a two user half-duplex cooperative communication system under the assumption of perfectchannel state information at the transmitters and the receiver. An achievable rateregion for the proposed method is characterized, and evaluated by computersimulation.iv
Cluster based sensors scheduling in a target tracking application with particle filter method
Circuit models with mixed lumped and distributed elements for passive one-port devices
In this thesis, to model measured data obtained from an actual passive one-port device, a circuit modeling method with mixed lumped and distributed elements is proposed. Namely, measured data is modelled by means of its Darlington equivalent, in other words, as a lossless two-port terminated with a resistance. Two network topologies are examined. The first topology is ladder networks connected with unit elements and the second one is cascaded separate lumped and distributed networks. In the proposed modeling method, analytic expression of the input reflection coefficient of the two-port model is obtained by using gradient method, and then, after synthesizing this two-variable function, the model is reached. Thus, for the first time in the literature, a two-variable circuit modeling method is presented.
New approaches for channel estimation in wireless communications
This thesis first proposes a computationally efficient, pilot-aided linear minimum mean square error (MMSE) batch channel estimation algorithm for the orthogonal frequency division multiplexing OFDM systems in unknown wireless fading channels. The batch linear MMSE will be converted to the sequential linear MMSE estimator due to fast convergence property and the simple structure. In addition to OFDM systems, focusing on transmit diversity (OFDM) transmission through frequency selective channels, this thesis pursues a channel estimation approach in time-domain for both space-frequency OFDM (SF-OFDM) and space-time OFDM (ST-OFDM) systems. This thesis also proposes a computationally efficient, pilot-aided linear minimum mean square error (MMSE) time domain batch channel estimation algorithm for OFDM systems with transmit diversity in unknown wireless fading channels. The proposed batch approaches (with or without transmit diversity) employ a convenient representation of the discrete multipath fading channel based on the Karhunen Loeve (KL) orthogonal expansion and finds MMSE estimates of the uncorrelated KL series expansion coefficients. Based on such an expansion, no matrix inversion is required in the proposed MMSE estimators. Moreover, optimal rank reduction is achieved by exploiting the optimal truncation property of the KL expansion resulting in a smaller computational load on the estimation algorithm. The performance of the proposed approaches is studied through analytical and experimental results. Then, in order to explore the performances, the stochastic Cramer-Rao lower bounds are considered for the proposed approaches. The effect of a modeling mismatch on the performances of the estimator is also analyzed. In order to explore the performance of the transmit diversity OFDM systems, the closed-form expression for the average symbol error rate (SER) probability is also derived for the maximum ratio receive combiner (MRRC) in these systems. Simulation results confirm our theoretical analysis, and illustrate that the proposed channel estimation algorithms for OFDM systems with and without transmit diversity are capable of tracking fast fading and frequency selective fading, respectively and improving overall performance.
Insertion loss design of RF filters with mixed lumped-distributed realization
In this study, the purpose was to develop a user friendly software tool which enables to design microwave filters with lumped circuit elements, distributed transmission lines and mixed lumped-distributed circuit elements. For the design tool, modern insertion loss approach is preferred because of several advantages such as flexible filter specification and easy control of filter characteristic. The work is mainly concentrated on low pass Butterworth and Chebyshev type filter designs. For the lumped element and distributed element filter designs the available insertion loss methods are explained and implemented. Solution to filter problems with lumped elements alone is well established in literature. However at microwave frequencies use of lumped elements or distributed elements alone in the circuit realization has serious implementation problems. Thus mixed lumped and distributed filter design have several advantages and flexibilities in monolithic integrated circuit layouts. Unfortunately, designing filters with both lumped and distributed elements has not been solved analytically yet. Beacuse of this reason, in this work a new approximation scheme is proposed to construct filters with lumped and distributed elements. Our new proposed approximation for mixed lumped-distributed filter design and microwave filter design tool will provide new possibilities and flexibilities in designing microwave filters.
Implementation of cooperative diversity in space frequency coded OFDM systems
Wireless technology is becoming an important part of our daily life. Although earlier wireless technologies were developed to provide voice transmission which did not require large amount of bandwidth, we are now equipped with intelligent mobiles, PDAs and laptop computers each having various applications connecting us to multimedia services that requires high bandwidth utilization. Technology world is converging to IP based systems where reliability and capacity plays an important role. For example the delay periods which can be neglected in voice transmissions can be crucial in services like on demand TV where data transmission have to be smooth. High performance and high reliability requires the bounds of capacity and coding structures to be increased. In this thesis we consider strong coding algorithms and diversity techniques in broadband channels. Space frequency codes which have been proven to achieve high performance in MIMO systems have been employed to single antenna systems by using cooperative diversity where deployment of multiple antennas is difficult. By constructing such a structure we target to achieve available diversities in the channel to improve the performance.
Optimization of integer-pel motion estimation for H.264 video encoding on TMS 320C6416T digital signal processor
Video processing is used in many applications such as broadcast television and home entertainments. Video applications have been revolutionized by the advent of digital TV and DVD-video players. The standardization of video compression technology is essential for many video applications. Today the state-of-the-art compression standard is the H.264 standard. In this thesis, an H.264 encoder implementation is optimized on Texas Instruments TMS320C6416T board for real-time processing. C6416 is a high performance and a low cost digital signal processor (DSP) chip that can achieve real time implementation of the algorithm. Thus, we choose C6416 because of based on our analysis of performance and cost. In this thesis, hierarchal motion estimation module is implemented for the H.264 encoder. First of all algorithm code was written in C language. Then performance critical parts are written in assembly. The resulting code is an optimized implementation on the Texas Instruments TMS320C6416T DSP. Simulations on TMS320C6416T reveal that the encoder processes 39-65 CIF frames per second, which satisfies 25 fps requirement for real-time applications.
Artifact reduction in occluded regions for motion compensated frame interpolation
In this paper, we propose a new method to be used in occluded regions for motion compensated frame interpolation. Frame rate up conversion is becoming much more important when usage of digital video systems increase. Standard conversion techniques are not effective in increasing the temporal resolution of video without any degradation. Therefore, motion compensated post processing techniques are developed to obtain the best possible image at the receiver. Specifying the correct or reliable motion vectors is the most important issue to be able to prevent the artifacts and deformed structures in interpolated frames. There are many methods in literature developed for that purpose. However, these methods generally create annoying artifacts in interpolated frames. Significantly, they can not provide acceptable results for covered and uncovered regions. In our work, we build upon the successful multi-stage post processing method which is used to obtain the correct vector for motion compensated frame interpolation. According to our analysis, all these post processing stages are effective to specify the reliable motion vectors, but they are not sufficient to solve the problem in covered and uncovered regions. In thesis, we suggest methods to be able to solve the problem occurring in these regions. In most of the adaptive methods, either past or future frames are used for interpolation in covered/uncovered regions, respectively, but in our proposals we interpolate the new frame by using a weighted combination of the four neighboring frames to minimize the sharp blocking artifacts that occur because of uni-directional interpolation in occluded regions. Experimental results show that our proposed method is robust in occluded areas, and reduces the sharp artifacts in the interpolated frame for covered and uncovered regions, and provides better picture quality.
Construction of multivariable broadband matching networks: An integrated design tool
One of the major problems in high frequency applications is designing matching circuits. Real frequency design of broadband matching networks is a commonly preferred approach since experimental termination data is directly utilized.In this work, design of matching circuits with real frequency techniques was examined. In order to design mixed circuits with lumped and distributed components, an integrated design tool was implemented by combining parametric and scattering approaches.This new approach was implemented as a MATLAB based design tool. This tool was integrated with .NET and a standalone, redistributable, user friendly general purpose matching circuit design tool was created.
True motion estimation for video frame rate up-conversion
Since the invention of high definition display technologies, video standards con-version problem has become an important task to store, transmit and displaynew video formats. One of the standards conversion issue, video frame rateup-conversion(FRUC) should be considered as an important problem for highdefinition displays because these displays reach high refresh rates up to 100Hzand low video frame rates should be increased before displaying. The solutionsgenerated by conventional FRUC techniques cause artifacts such as jerky motionand judder effect on the screen. The objective of this thesis is to remove judderand occlusion artifacts and generate a smooth object motion for high definitiondisplays. In order to solve the problem, this thesis describes a multi-stage motionvector (MV) post-processing technique for true motion estimation and proposesa new video frame rate up-conversion method with occlusion adaptive overlappedblock motion compensation, which increases the temporal resolution without lossof spatial resolution and consistency. Conventional block matching algorithmsyields us unreliable MVs which should be post-processed in the refinement stagesto get a consistent MV field in spatial domain . Relations between neighboringMVs are utilized to detect occlusion regions. Overlapped Block Motion Compen-sation (OBMC) is adapted to these problematic regions in order to reduce haloartifacts caused by occlusion. Compared to existing methods, the proposed algo-rithm achieves FRUC with reduced loss of spatial resolution and reduced amountof artifacts.
Wavelet based image coding and interpolation
The need of higher resolution on imaging systems and displays increases everyday.Data rates and bandwidth are still limited to satisfy the demands of enhanced resolutions.So, we should develop intelligent enhancement tools which yield higherresolution images with comparably limited bitrate. In this thesis, we examine thequality enhancement capabilities of two dierent approaches, i.e. image codingand image interpolation. Better image coding algorithms are capable of producingenhanced higher detail images at the same bitrate. For that purpose, wedesign an ecient and intelligent wavelet based image coding algorithm that codesthe hierarchical description of wavelet coecients instead of coding themselves.Namely, we introduce the hierarchical quantization index tree which is composedof quantization index classes. These index classes are constructed using combinationof similar wavelet coecients which leads an adaptive structure. Then,this hierarchical tree is optimized by a simple rate-distortion analysis to achieveecient bit allocation among various dierent regions of natural images. In thesecond part of the thesis, we propose a wavelet based interpolation algorithm thatexploits the correlation between high resolution(HR) and low resolution(LR) images.Basically, we design linear minimum mean square error lters between HRand LR images to recover lost high frequency information. For modeling the relationshipbetween two resolutions, we use two dierent approaches, i.e. blockbased lter design and context adaptive lter design. In block based lter designalgorithm, we partition the image into blocks to capture local frequency variations.In context adaptive algorithm, we use a simple context to adapt dierentimage structures. We also briey mention how these new approaches can beintegrated into a novel coding + resolution enhancement joint framework. Simulationsshow that both our coding and interpolation algorithms perform betterthan most existing schemes.
Ring array processing for forward-looking intravascular and intracardiac ultrasonic imaging
Forward-looking (FL) catheter-based imaging systems are highly desirable for guiding interventions in IVUS and ICE applications. One of the main challenges in the array based FL-IVUS systems is the large channel count which results in increased system complexity. Synthetic phased array processing with reduced firing count simplifies the front-end and hence can enable 3-D real-time imaging. Recently, we have investigated dual-ring arrays suitable for IVUS and ICE imaging. In this dissertation, we present and explored two different optimized array designs based on dual circular and hexagonal rings that are suitable for synthetic phased array processing with reduced number of firings. We used simulated annealing to optimize the wideband PSF of the array system. The optimal firing set from different Tx-Rx elements eliminates most of the redundant spatial frequencies in the coarray and produces low side lobes and number of firings. The optimization procedure starts with the calculated wideband PSF of the full set coarray including all Tx-Rx firing combinations. In every iteration step, a predefined coarray with only one element change is compared with the previous set in terms of cost function. For the two designs, we obtained optimal reduced set by minimizing both the peak side lobe level and the integrated side lobe energy. For two test cases, we obtained reduced sets of 512 and 256 coarray elements, and constructed 2-D PSFs. The simulation results show that the optimized dual circular and hexagonal ring arrays provide up to 10-dB and 5-dB improvements, respectively, in peak near side lobe level with no widening in main lobe width. Our test results show that the side lobe levels of the dual circular and hexagonal arrays can be remarkably and controllably reduced by using simulated annealing optimization, and the resulting optimal array designs can meet the resolution and speed requirements of 3-D real time IVUS and ICE imaging.
Solid-state arrays and beamformers for side-looking intravascular ultrasonic imaging
Current solid-state intravascular ultrasound (IVUS) devices use a one-dimensional(1-D) rectangular array wrapped around the cylindrical catheter. Beamforming inthese devices is limited to synthetic phased array processing, and the image quality isdegraded by low signal to noise ratio (SNR) and lack of focusing in elevation. Wepropose new solid-state reconfigurable cylindrical array configurations, based onnon-uniform aperture sampling, enabling realization of low-cost integrated phasedarray beamformers based on the latest IC and transducer integration techniques. Weemployed non-uniform (Fresnel) aperture sampling with identical differential delaysbetween adjacent elements to reduce dynamic beamformer complexity. For thispurpose, the angles of the element boundaries and centers are determined byequalizing path difference of neighboring elements. For Side Looking-IVUS (SLIVUS),the image plane is scanned through stepping the active aperture, and theactive array must be reformed in each step. Fresnel sampled ring array leads to asimplified phased array beamformer with N/2 identical differential delays. The samelogic applies to a two-dimensional (2-D) array with NL×NE-elements over acylindrical patch. We compared the proposed and existing array designs throughnumerical point spread function (PSF) simulations. The uniform and Fresnel sampled1-D ring arrays perform similarly in terms of beam quality while the latter improvesimage SNR by a factor of ?NL due to synthetic vs full phased array processing. TheFresnel sampled 2-D array performs dramatically better in elevation compared to 1-Dring array. The additional contribution of 2-D array on image quality is high SNRproduced by phased array beamforming. The proposed Fresnel sampling ofcylindrical aperture simplifies phased array beamformer by halving the requireddelay element count and transforms the complex 2-D beamformer into a separablebeamformer that requires simple 1-D beamformers. The resulting 1-D and 2-D SLIVUSarray configurations could be realizable using monolithic integration of theanalog beamformers with transducer arrays.
A cooperative communication approach to cognitive radio
Cognitive radio, whose main idea is to have intelligent devices utilize the limited resources, such as frequency spectrum in wireless channels, is becoming a groundbreaking technology. The aim of this thesis is to apply the cooperative communication techniques to the systems which consist of cognitive radio terminals and to develop new cooperative encoding and decoding protocols to reach higher data rates. Accordingly, the performance of a cognitive system where there is one primary (licensed) user and one secondary (unlicensed) user in a network is analyzed considering two approaches: maximization of the sum rate, and maximization of the rates achievable of the secondary user. Optimal power allocation policies for the cooperative cognitive multiple access channel (MAC) are obtained, and the resulting power distributions and achievable rate regions are computed via simulations.
Compress and forward based cooperation strategies in wireless networks
Cooperative communications are emerging as an alternative technology to MIMO communications. In this thesis, we first summarize information theoretical analysis of known channel types and also consider three fundamental relaying techniques known as Amplify and Forward, Decode and Forward and Compress and Forward. We propose two compress-and forward based cooperation protocols for a two-user multiple access channel, and obtain their achievable rate regions. The performance of the proposed models are is compared with the performance of other two-user cooperative protocols proposed in literature.
Deniz suyu parametrelerinin MEMS tenolojisi ile ölçülmesi
The system is used to measure the pressure and the temperature values up to hundred meters of the sea water. This project is also called as expendable environmental probe. The probe contains a temperature sensor, a pressure sensor, and system electronics. It is needed for the calculation of the performance of the sonar of warships, oceanographic institutes and oceanographic purposes.The probe include the sensors which developed by using Micro Electro Mechanical Systems technology. The probe is an expendable probe capable of reaching over hundred metres of depth when dropped from a moving ship, and a thin conducting wire is transferred through the data to the ship. This data is collected, stored, and displayed the data acquistion system.Data was collected from the temperature and the pressure sensors but the data could not displayed the data acqusition system with one wire system. One wire system is used only a few meters but I want to communicate the system up to hundred meters. Therefore I decided to use two wire system with communication module. In conclusion, the system was prepared in conformance with hardware design and the communication system was changed with two wire system design.
Cooperative strategies, achievable rates and resource allocation for OFDMA channels
The design of next generation wireless communication systems brings along new challenges, since the degrading factors such as fading and multi-user interference become harder to deal with as the number of users and the bandwidth requirements increase. Orthogonal Frequency Division Multiple Access (OFDMA) is a multiple accessing technique which provides a solution to both of the problems above: it provides a relatively simple way of assigning available bandwidth to users, while avoiding interference; and at the same time, it converts a frequency selective fading channel, to parallel flat fading subchannels, hence reducing the effects of intersymbol interference. However, in wireless channels, what is traditionally considered as interference is in fact side information, and combined with the diversity created by the orthogonal subchannels in OFDMA, this side information can be carefully taken advantage of to increase the rates achievable by the users. In our thesis, without imposing any prior constraints on subchannel allocation, we investigate cooperation strategies, achievable rates and resource allocation for OFDMA channels.We propose new cooperative encoding strategies for wireless communication networks over OFDMA channels. We particularly focus on a two user cooperative OFDMA system, based on block Markov superposition encoding (BMSE). We obtain expressions of the resulting achievable rate regions for all proposed cooperative encoding strategies. We show that, by allowing for re-partitioning and re-encoding of the cooperative messages across subchannels, it is possible to better exploit the diversity created by OFDMA, and higher rates can be achieved.In order to take full advantage of the diversity created by OFDMA, we then introduce a channel adaptive cooperation strategy for OFDMA, and optimize the transmit powers as a function of the channel states. We provide the optimality conditions that need to be satisfied by the powers associated with the users? codewords and derive the closed form expressions for the optimal powers. We propose two algorithms that can be used to optimize the powers to achieve any desired rate point on the rate region boundary. We observe that, utilization of power control to take advantage of the diversity offered by the cooperative OFDMA system, not only leads to a remarkable improvement in achievable rates, but also may help determine how the subchannels have to be instantaneously allocated to various tasks in cooperation.
Automatic speech recognition system for Turkish spoken language
The transmission and storage of speech sounds is possible for decades. In addition by usingsignal processing techniques, it is also possible to process speech signals. By using timeand frequency analysis of speech signal and several machine learning algorithms, it ispossible to build a system which is used to recognize spoken words. Such systems arecalled Automatic Speech Recognition systems.In our work, we have used the Automatic Speech Recognition system for Turkish spokenlanguage which has built by BUSIM speech group. However, the output of the recognizer isthe list of spoken words. Even for humans it is a very hard task to understand a text withoutpunctuation symbols. Hence to build more complex recognizer whose goal to perform topicsegmentation and topic summarization, the output of ASR should be divided into sentencesat first.Our goal is to build a system which performs the sentence segmentation. In our work wehave used ASR system to obtain word level and phoneme level time marks and by usingthat time marks with the audio files, we have extracted prosodic features, where theprosodic properties of speech contains information about the punctuation in the text, whichis not available at the output of ASR system.
Video quality assessment using subjective and objective metrics
Since past few years, quality assessment of digital videos has acquired a lot of attention in the video processing community, leading to the growth of different 2D and 3D objective metrics for images and videos. Some of these metrics correlate quite well with the Human Visual System (HVS), while some do not. Most of the popular objective metrics do not perfectly correlate with the HVS. This correlation can be measured by the opinion scores based on observations of human subjects. In this thesis, we extensively study about Video Quality Assessment and complexity of HVS by proposing a method which helps us to find the best correlation between HVS and the objective metrics. The motive behind this work is to to introduce an objective metric that is adequate to predict the MOS (Mean Opinion Score) of distorted video sequences based on the FR (Full Reference) method, keeping in mind the benchmark set by the Video Quality Expert Group (VQEG). Subjective quality evaluation experiments using the human observers were performed, that is, a group of non-experts viewing the video sequences in original and distorted forms, as it is the human perception method which is evidently the most reliable one. Thereafter, objective test results, including the introduced metric, called PQM (Perceptual Quality Metric), and several other state-of-art metrics, are compared with MOS provided by human subjects. These comparisons elucidate the level of correlation between tested objective metrics and MOS. As an ultimate objective, we show the accuracy and monotonicity of various quality metrics, including PQM. The results obtained show that the PQM is better correlated with the human subjective judgement.
Partner selection and resource allocation in single-cell, multi-cell and cognitive cooperative multiple access channels
Wireless communication has been widely used across the globe for several years. As the technology advances and gains popularity, more people start using it and more throughput is needed. For that reason, several techniques are being researched such as MIMO systems and cooperative networks. Wireless cooperative networks make use of the natural property of radiation of electromagnetic waves. Since the waves emitted by a user in the network can be heard by all users in the network, one cooperating partner uses this overheard information to increase throughput. However, to maximize the throughput of a system, cooperating partners must be selected intelligently. In this dissertation, first, we will summarize cooperative communication basics, techniques we used in convex optimization and graph theory. Then, we will show how these concepts can be used together to optimally maximize system throughput and propose lower complexity yet nearly-optimal partner selection algorithm will be proposed. Under the light of the results of this work, in the next chapters, we will introduce a novel fractional frequency reuse scheme which encourages users to cooperate and allow system to support more users. In the last chapter, a cognitive scenario will be used for one cell and we will present the optimal partner selection scheme for system throughput maximization.
Ground plane detection using an RGB-D camera for quadcopter landing
The purpose of this study is to build an autonomous quadcopter which is capable of automated detection of landing zones and landing. To ensure this there are important steps linked together. The first step is to build a quadcopter that can fly. We have used a commercially available quadcopter platform and an Ardu Pilot Mega control card to build a low cost, easy to implement, and stable platform. An open source firmware, named Arducopter, is used for the control card. This system can take-off, land, hover, and follow a given flight path. In order to detect of landing zones and safe landing, we propose to use an RGB-D camera as a sensor and a small onboard pc as the computing engine. Hence, we have modified the acquired quadcopter frame to integrate additional components. In this thesis, we propose a novel and robust ground plane and obstacle detection algorithm based on depth information using RGB-D camera. Moreover, our method was compared with V-disparity algorithm from the literature. It has been shown that our algorithm performs better than V-disparity method and produces useful ground plane-obstacle segmentations, even for difficult cases. The method is able to work in highly dynamic platforms. This algorithm is generic in the sense that it can be used for different forward-facing RGB-D placements, for example in ground vehicles or robots. Moreover, we developed a pre-process to allow the use of the method for down-facing sensor view positions as the core method is inadequate for landing zone detection. The proposed method compensates the movements of the camera caused by the air vehicle, and detects the ground plane obstacles successfully. It has been shown that the use of RGB-D camera allows ground plane and landing zone detection even in no-light conditions. All necessary components in this thesis were financed by FMV Işık University internal research funds BAP-10B302 project.
Design and optimization of multiuser cooperative protocols for wireless networks
Recently, the main trend in wireless communications has shifted from voice transmission to data communication. Naturally, this shift caused an increase in the data rate and bandwidth requirements. Being a limited and thus expensive resource, wireless spectrum needs to be used efficiently. For higher spectral and data rate efficiency, new spectrum-allocation policies as well as new transmission techniques are needed. Cooperative and cognitive radios are promising emerging technologies, which can enable efficient spectrum resource utilization as well as high data rate transmission in the next generation wireless networks. This thesis addresses wireless relay networks consisting of multiple cooperative and/or cognitive nodes. The main contribution of this thesis is the extension of cooperative strategies from two users to three user settings, focusing on mutual cooperation. In the first part of this dissertation, we concentrate on a superposition block Markov encoding based three user cooperative communication scheme for a fading Gaussian multiple access channel. We consider all possible channel conditions between users and propose a channel adaptive block Markov encoding strategy. In the second part of this thesis, for the same three user MAC model we discuss a new channel non-adaptive superposition block Markov encoding structure which enables all three users to cooperate collectively as well as in pairs. The proposed cooperation models not only provide increased diversity to all participating users, but also contain as special cases the multiple relay channel and the multiple access relay channel. In the third part, we focus on the joint use of cognition and cooperation. In particular, we consider overlay and underlay cooperative-cognitive radio models with one primary and two secondary users. We extend the cooperative encoding models developed for the three user MAC to the cognitive set-up. In all three problems studied in this thesis, we obtain new and improved achievable rate regions, by cognition and/or cooperation.
Fingertip ECG signal based biometric recognition system
The idea is the; realize biometric recognition system by using ECG signal which was began to use in the last 10 years. Until now, ECG based biometric systems have been developed by using the database which ECG signals are taken from the subjects' chest by using patient monitor or high speed data acquisition systems. For constructing database, most researchers have used three disposable ECG electrodes on subjects' chest to extract the ECG signal. Because of that, ECG based biometric systems were being considered hard to use. For this reason, we want to make a system that is easy to carry, and easy to apply on subjects. In most of biometric systems, ECG database have constructed by using ECG signal of the subjects that were taken by using electrodes which were located in left and right side of the heart with a reference electrode on right leg. However, in our system, the database consisted of the ECG signals which were taken by using patient's right and left thumb. The difference of our system with the others is; there is EMG noise which is in the same frequency range with ECG signal. Because of frequencies are the same, it is very hard to eliminate with filters. For this reason, database was enhanced by applying the different filters on ECG signal to reduce the noises for higher recognition rate. First week of the dataset -which was obtained by using ECG signals of 30 people in two separate weeks- is extracted the personality information by performing AC/DCT and MFCC methods and is reserved as training dataset. ECG signals which obtained by AC/DCT methods in second week are called as test dataset. First candidate is determined by putting the AC/DCT features of an unknown person into the LDA classifier. In the meantime, same person's MFCC features put into the LDA classifier and the second candidate is determined. If these two candidates are the same, they are labeled as A and B person. If they are not the same person, then QRS frames of the proximate two candidates obtained from AC/DCT features and QRS frames of the proximate two candidates obtained from MFCC features are sent to K-NN algorithm. QRS frames of these 4 candidates are ascending according to the proximity to the QRS frame of unknown person, and nearest candidate to unknown QRS segment is labeled as A and B person. Proposed method was reached to success at rate of %96 average frame recognition.
Energy and data cooperation in energy harvesting multiple access channels
A typical wireless device is limited by its fi nite battery, and energy limitation is as a critical bottleneck on wireless network performance. In addition to the EM radiation caused by high power wireless devices, the batteries used in those devices need to be recharged frequently, causing fast depletion of earth's resources; and are discarded after a while, causing environmental pollution. Moreover, in some scenarios, such as sensor networks deployed in rural areas, it is simply not feasible to replace the batteries of the wireless nodes, hence using nodes which can replenish their energy is of paramount importance in prolonging the network lifetime. As a result, the concept of green communication and the idea of designing communication protocols based on energy harvesting constraints of individual nodes have recently received tremendous attention. With this motivation in mind, this thesis is devoted to designing cooperation and transmit scheduling techniques for networks powered soleley by harvested energy. We study energy cooperation in a two user multiple access channel and a two user cooperative multiple access channel with and without battery limitations. We characterize, and optimize the departure region for a two user multiple access channel, under six di fferent models, with varying levels of cooperation and energy storage capabilities. We aim to find the optimum energy management policies of the transmitters to maximize the achievable departure region over a fi nite transmission duration.
Compression of the mammography images using quadtree based energy and pattern blocks
Medical images, like any other digital data, require compression in order to reduce disk space needed for storage and time needed for transmission. This thesis offers , a novel image compression method based on generation of the so-called classified energy and pattern blocks (CEPB) is introduced and evaluation results are presented. The CEPB is constructed using the training images and then located at both the transmitter and receiver sides of the communication system. Then the energy and pattern blocks of input images to be reconstructed are determined by the same way in the construction of the CEPB. This process is also associated with a matching procedure to determine the index numbers of the classified energy and pattern blocks in the CEPB which best represents (matches) the energy and pattern blocks of the input images. Encoding parameters are block scaling coefficient and index numbers of energy and pattern blocks determined for each block of the input images. These parameters are sent from the transmitter part to the receiver part and the classified energy and pattern blocks associated with the index numbers are pulled from the CEPB. Moreover, in the second part of our method we used Quadtree too. By this way, all CEPB from quadtree results determined for each block of the input images too. input image is reconstructed block by block in the receiver part using a mathematical model that is proposed by 2 different method: • Reconstruct Based on one block size • Reconstruct Based on Quadtree Evaluation results show that the method provides considerable image compression ratios and image quality even at low bit rates. Test result have shown that Compression ratio and PSNR results is acceptable, moreover, Quadtree method gives better results that fix based block size.
Biometric recognition using bio-signals
The main objective of the project is to increase the recognition rate by establishing a multimodal biometric recognition system that uses two different biometric characteristics, as bio-signals. Today, institutions use biometric recognition systems quite often to provide security for many areas such as information security and physical security. The importance of these systems increases day by day in the direction of technological development and increasing demand. Recognition systems based on biometric characteristics are more reliable, because of the possibility of forgetting or losing knowledge in the recognition systems based on knowledge (eg: password) and the possibility of being stolen or guessed by third persons in the biometric recognition based on possessed (eg: card). However, the fraud techniques are also developing in the direction of technological developments and biometric characteristics cannot be renewed in case of imitation, hence the use of multiple biometrics recognition system may be a solution to this problem. At the same time, the use of multiple biometrics increases in the security of systems. In this thesis, a biometric recognition system, which uses the Electrocardiogram (ECG) and speech signals of the person, was created. Since there was not enough time and possibility, an artificial database was generated with obtaining these signals from various sources. First, the MIT-BIH Arrhythmia Database was used for ECG signals. This database consists of 48 ECG signals, which belong to 22 females and 26 males. In accordance with this database, a database was created for the speech signals, which were obtained from the website, given in [1]. The features of the biometric signals were extracted by AC / DCT (Autocorrelation/ Discrete Cosine Transform) method for ECG signals and by Mel Frequency Cepstrum Coefficients (MFCCs) method for speech signals. The data, which were obtained from the feature extraction, were then classified by the Gaussian Mixture Model (GMM) method. The scores, which were obtained from the classification process, were fused as a single individual's data, and the decision-making step was passed. Recognition rates were obtained in the decision making step. The recognition rate for the ECG signal was %87.50 and 42 persons were matched correctly. The recognition rate for 2 seconds speech signals was %58.33 and 28 persons were matched correctly. Normalization was applied before the fusion of these two datasets. The recognition rate after the fusion was %70.83 and 34 persons were matched correctly. However, when the recognition rates are considered, it has been observed that the recognition rate, which obtained after the fusion, is lower than recognition rate of the ECG signals. Therefore, instead of 2 seconds speech signals, 10 seconds speech signals were used. In this case, the recognition rate of the speech signals was %97.9 and 47 persons were matched correctly. Then, normalization was applied again and two datasets were fused. After the fusion, the rate of recognition reached %95.8 and 46 persons were matched correctly.
Evalutation of feature selection and encoding methods for superpixel image parsing
This thesis is about image parsing which is one of the important problems in computer vision. The goal of image parsing is segmentation of object and labeling of each object. Recently, a popular way of image segmentation and classi cation is superpixels. Image is segmented into visually logical small regions by using superpixel algorithm and then, superpixels are parsed into di erent classes. Classi cation performance is signi cantly a ected by the properties of superpixel algorithm and parametric settings. SuperParsing is one of the superpixel-based image parsing algorithm and provides a succesful nonparametric solution for image segmentation and classi cation problem without any need for classi er training. SuperParsing labels each superpixel based on feature matching between the superpixel and a subset of the training superpixels. The training subset is determined by global matching between the test image and the training set. For superpixel matching the method makes use of a rich set of superpixel features. Class conditional log-likelihood is computed based on these matched features. The main objective of this thesis is to show improvements in labeling accuracy percentage by using feature encoding and selection methods, including learned features from Convolutional Neural Network (CNN) models. We perform two different encoding methods to selected features of superpixels and show that feature encoding improves parsing accuracy. The applied feature encoding methods are locality-constrained linear encoding (LLC) and kernel codebook encoding (KCB). LLC encoding method gives us 2:6% improvement on per-pixel accuracy for SIFT Flow dataset and 6:8% improvement on per-pixel accuracy for 19-class LabelMe dataset. KCB encoding method gives us 3:6% improvement on per-pixel accuracy for SIFT Flow dataset and 6:2% improvement on per-pixel accuracy for 19-class LabelMe dataset. All these results are overall improvement which are computed over original SuperParsing. Most recent studies about image segmentation and classi cation use CNN to improve their accuracy percentage. Features extracted from pre-trained networks, which are trained on large image databases, can be used in addition to handcrafted features in image segmentation. Last layer of these CNN models give the best features for classi cation. We test learned CNN features together with KCB or LLC encoding methods. We use CNN features both for global matching and superpixel matching. These tests give us 7:3% overall improvement over original SuperParsing on SIFT Flow dataset and 10:3% overall improvement over original SuperParsing on 19-class LabelMe dataset.
Data compression for smart grid infrastructure
Embedded systems, which work with large data sets, need larger memory and consume much more data traffic when compared to other IOT (Interne Of Things) devices. In this study, Hu man's Lossless Compression Algorithm is optimized and implemented on a Electric Meter Reader Modem (DR-502-DCL). Compression Algorithm used for storing the data which is measured by the electric meter. Huffman's Algorithm optimized for memory (RAM) on a multi-threaded embedded system. STM32F412 microcontroller and Winbond 128mbit embedded fash memory is used for this study. Compression ratio of %59 is reached with eliminating repetitive character groups.
Co-training using prosodic, lexical and morphological information for automatic sentence segmentation of Turkish spoken language
Sentence segmentation of speech aims to detect sentence boundaries in a stream of words output by the speech recognizer. Sentence segmentation is a preliminary step toward speech understanding. It is of particular importance for speech related applications, as most of the further processing steps; such as parsing, machine translation, and information extraction, assume the presence of sentence boundaries. Typically, statistical methods require a huge amount of manually labeled data, which is time and labor consuming process to prepare. In this work, novel multiview semi-supervised learning strategies for the solution of sentence segmentation problem are proposed. The aim of this work is to nd e ective semi-supervised machine learning strategies when only a small set of sentence boundary labeled data is available. This work proposes three-view co-training and committee-based strategies incorporating with the agreement, disagreement and self-combined strategies using lexical, morphological and prosodic information, and investigates the performance of the proposed learning strategies against the baseline, self-training, and co-training. The experimental results show that the proposed learning strategies highly improve the sentence segmentation problem since data sets can be represented by three redundantly su cient and disjoint feature sets.
Deep learning techniques for building density estimation from remotely sensed imagery
This thesis is about point-wise estimation of building density on the remote sensing optical imageries by applying deep learning methods. The goal of the project is to reduce mean square error of the estimated density by applying architectural modifications on the deep learning network and using augmented training data. Our results show that fast and accurate building density estimation is possible by using vanilla CNNs. Sigmoid layer addition, simplification of the network for small dataset and data augmentation improves accuracy in the regression. Data augmentation is the most effective method to reduce RMSE in this thesis.
Pil test ve yönetim sistemi
This thesis is about meeting and answering the needs of devices with batteries for battery testing and management. The goal of the battery test and management's is to use the full capacity of the device and to be able to see the need for changing the battery at the right time. The increase in the number of electric appliances and electric vehicles also led an increase in the need of batteries recently. That situation affects every aspect of the daily life and healthy performance of batteries became even more important in terms of environmental security, effciency and so on. Engineers have developed several solutions and technologies for these problems. Open-Circuit Voltage Measurement Method, Coulomb Counting Model and Kalman Filters are some examples. On the other hand State of Charge method helps us to get information about the health of the battery, However, all applications for the problem so far have been quite huge, inaccessible and hard to use. Since the batteries became an important of daily life more than ever, these applications are needed to be accessible, economic, precise and user-friendly. The aim of the thesis is providing an affordable and user-friendly device by integrating mainly the Coulomb Counting Method with several others successfully. Using Raspberry in the device helps it to be accessible, economic, fast and adaptable to dfferent interfaces. All the components that had been used in the device in this thesis have approximately from %1 to %5 error term and therefore quite precise about the estimations. Today, the cost and the complexity of Battery Test and Management systems led waste of millions of batteries through inffective usages. That harms the environment and economy of the world. The device that had been developed in this thesis, provides accessible, affordable with its 80$ price and user-friendly solutions for battery testing and management problems.
Metamutator: Its realizations and its applications
Mutators became very popular after 1971, when Leon Chua realized the memristor, his postulated fourth circuit element, using them. The reason for the popularity of mutators, which are easily realizable 2-port devices, lies in the fact that they render possible the simulation/emulation of, hence experimentation with, postulated non-existing (not off the shelf available) elements like memstors . On the other hand, in the literature when simulating/emulating elements with 2-ports, many 4-port "generalized mutator-like" realizations that nobody has been able to identify are being used. These underlying 4-ports, and their ability to act as a mutator when two of the ports are properly terminated, have been thus named metamutator. In this thesis, in addition to introducing some of metamutator realizations, newly designed metamutator circuits with one or two active devices are introduced. Also, a new active device, with only twelve transistors, named Additive and Differential IC (AD-IC), is proposed with its layout and use in metamutator circuit design. In addition, many 1-port and 2-port circuit realizations using metamutators have been introduced. 1-port applications are memstor simulation/emulation, floating and/or grounded impedance scaling which comprise inductance simulation, capacitance multiplication, oscillators and Frequency Dependent Negative Resistor (FDNR) simulation. 2- port applications cover Voltage Mode Multiple Input Single Output (VM-MISO) and Current Mode Single Input Multiple Output (CM-SIMO) universal filters and implementations of transconductance and transimpedance amplifiers. Also, two different applications of AD-IC: AD-IC based analog multiplier and AD-IC based full-wave rectifier are proposed in the thesis.
Multiband and concurrent matching network design
Multiband network design has become a big issue due to the current improvements in RF and MW technologies. Since the new wireless protocols come up fast such as 3G, 4G, 5G etc., the RF and MW hardware must be designed to provide services to all distinct technologies in simultaneous regimes. To avoid interference, noise, harmonics, etc., the use of multiband matching or filtering at the front-end or back-end of RF/MW devices is an inevitable work. Particularly multiband matching in a multiband system is always necessary to be sure of proper power transfer between each multiband units. In this thesis, the multiband matching network design problem is addressed under the analytical perspective. Therefore, a deep investigation on network theoretical foundations is performed. In addition, only passive networks are concerned for providing concurrent operation. One of the well-known semi-analytical network design methods, which is called real frequency technique, is taken as a main domain to develop the thesis contributions. Two main approaches are elaborated. The first approach is the generation of proper frequency mapping functions and determination of prototype networks to design multiband matching network. The proposed frequency transformation approach is useful and practical for multiband filter and matching network designs. As a second research track of the thesis; a synthesis based approach is aimed to be developed for generalized multiband matching network designs. A proper integration of the proposed approach with RFT matching techniques and examination of the implementation results are provided.
A new model for speech compression
This thesis aims to propose a new low bit-rate speech compression method. In the proposed method, classical SYMPES (Systematic Procedure for Predefined Envelope and Signature Sequences) based predefined signature and envelope sequences are obtained using zero-cross and phoneme-based segmentation. Some disadvantages of the traditional SYMPES technique, like computational complexity, relatively long encoding times, and so on, are also reduced in the new version in order to obtain lower bit rates. The new approach significantly reduces the bit-rate and yields high compression ratios with more intelligible speech quality than that of the classical SYMPES approach. Furthermore, in comparison to other traditional techniques such as the CELP (Code Excited Linear Predictive) coding algorithm, experimental findings demonstrate that at almost the equal bit rates, extremely promising speech quality is produced.
Fingertip electrocardiogram and speech signal based biometric recognition system
In this research work, we presented a one-dimensional CNN-based person identification system which depends on the combination of both speech and ECG modalities to improve the overall performance compared to traditional systems. The proposed method has two approach: one is to develop combination of text-independent speech and fingertip ECG fusion system, the other one is to develop a robust rejection algorithm to prevent unauthorized access to the fusion system. In addition to the system robustness, we have developed an ECG spike and inconsistent beats removing algorithm, which detect and remove the problems caused by either portable fingertip ECG devices or movements of the patients. First approach has been tested on 30, 45, 60, 75 and 90 people which were taken from LibriSpeech Corpus database and combination of both CYBHi and our private fingertip ECG database. The 3-fold cross validation test setup has been conducted while system working time was set to 10 seconds. In the first experiment, we achieved 90.22% accuracy rate for 90 people for ECG based system. For the speech based system, 97.94% accuracy rate has achieved for 90 people. For the combination of both system, 99.92% accuracy rate has been achieved. For the second approach, 90 people for ECG and Speech database were being used as genuine class, 26 people as imposter class, and after the performance evaluation in optimum rejection thresholds, 71.08% accuracy rate for imposters rejection and 71.05% accuracy rate for genuine recognition has achieved for ECG based system. For the speech based system, imposter class were 87.82% accurately rejected while genuine classes were 86.48% accurately identified. The combination of both system has achieved 91.68% accuracy for genuine identification rate whereas 96.05% accuracy for imposter rejection.
The behoavioral modeling and an application of SAR ADC
Analog to digital converters are the main interfaces between analog world and digital circuits. The quantities which are meaningful for human senses such as voice, light, temperature etc. are not suitable for processing directly by electronic circuits. The analog values must be transformed to their electrical representations, which means the analog values are expressed as magnitude, phase or frequency of current or voltage. Finally, this information are converted to digital bits via analog to digital converters in order to be processed by digital circuits. Similarly, the digital data which consists of generally logic-1 and logic-0 are not understandable for humans. Hence, the digital to analog converters transform the bit stream to analog representations. Therefore, the need for data converters never decreases. Various types of analog to digital converter topologies are developed according to system and area requirements. Low power consumption, small area usage and design simplicity helped, SAR ADCs to become the most popular architecture at moderate speed and moderate to high resolution application in the last decades. Even though the designed ADC has physical n-bit resolution, never achieved n-bit ENOB performance in real conditions. Capacitor and transistor both arising from fabrication process, deteriorate the performance of ADC directly. The value of the capacitors affects the transfer function on the first order. Hence, the capacitor mismatch is the main limitation for high resolution ADCs. External noises which originates from electrical network, power supplies, light sources and other environmental factors such as temperature, electrical/magnetic fields are unfortunately inevitable. Also there are different internal noise and error mechanisms arise from active and passive devices. The effects of these distorting factors can be reduced using special design and fabrication techniques, but never eliminated. The obtained performance has an upper limit, no matter how successful the design. If redundancy or digital calibration techniques are used, the performance can be improved. However the power consumption and area usage increase after the application of these techniques. The purpose of the thesis is to examine the limitation of SAR ADC with and without digital calibration and redundancy. In this thesis, a differential SAR ADC model is created (with and without redundancy) in MATLAB environment in order to investigate error mechanisms and non idealities. The effects of capacitive mismatch, parasitic capacitances of DAC array, comparator noise and offset, the noise of reference voltages, sampling noise and settling behavior on ADC performance can be determined using MATLAB model. Also the optimum configuration for DAC can be chosen and, if necessary, number of the thermometric bits can be specified. This model can also be adapted easily other SAR topologies. Although this model does not include some problems such as charge injection, clock feedthrough and kickback noise, it efficiently models SAR ADCs, reducing simulation times by multiple order of magnitudes. A 14 bit differential SAR ADC is design in Cadence design environment. The design is fabricated in TSMC 65nm CMOS technology. The active die area is $2mm^2$ with full rail to rail input swing of $5V_{pp}$. A 74 dB SNDR is achieved at 20M SPS sampling rate and 5 MHz input frequency. The total power consumption is less than 25 mW from 1.2 V and 2.5 V supply voltages. In conclusion, SAR ADC performance limitations and design bottlenecks was determined thanks to the MATLAB model.The realized behavioral model will be a good start point for a designer who is dealing with SAR ADC.
Wireless and batteryless instant velocity sensor for certain daily motions
The miniaturization studies in both integrated circuits (IC) and microelectromechanical systems (MEMS) have made it possible to have electronic systems that can operate with very low power. Sensors that are used for monitoring in the buildings, factories or military facilities and wireless sensor networks are examples of such systems. Most of these systems are powered up via batteries which may cause some problems such as replacement of the batteries or disposal of these batteries. Since these systems can operate with very low power, they can be powered up with miniature renewable energy sources. The energy sources that are used in such systems are commonly: solar, thermal, acoustical, mechanical and vibrational energy harvesting systems. Another main energy harvesting source is the ambient motion. It is possible to have a harvesting system, that utilizes human motion, which can power some small mobile equipment. In this thesis, a system is designed that harvest energy from human motion by electromagnetic induction to sense certain daily motions.
Design of high efficiency buck converters
Buck converters are used for battery operated systems which supply a wide current range and very low output voltage ripple even when the input voltage changes by more than two orders of magnitude. The converter has to sustain an excellent power efficiency and be able to support load current switching from 0A to Imax with minimum output voltage ripple. Increasing the power efficiency in a switching buck converter requires decreasing the power losses in both light and heavy load operations. However, the light load efficiency is the major concern since portable systems spend their majority of time in idle mode. This work introduces a novel method for increasing efficiency in buck converter systems by utilizing an adaptive output stage size selection technique which decides on the active number of stage segments by comparing capacitive and resistive losses. It takes into account the input supply voltage variation, operating frequency, gate capacitance, actual RON information and load current and finds optimal stage size to obtain optimal efficiency in a switching converter. This is the only work in the literature which conducts high efficiency in light loads without changing the modulation scheme or frequency of switching converter. Two patent applications have been made throughout this work as US14695492: Method for an Adaptive Transconductance Cell Utilizing Arithmetic Operations and US14712941: High Efficiency DC/DC Converter with Adaptive Output Stage.
Design of mixed-mode building blocks for an electronic stethoscope IC
Detecting respiratory issues early is crucial; listening to lung sounds can aid in early detection. However, traditional auscultation requires good listening skills and clinical experience. There are other ways to create images of the lung such as Magnetic Resonance Imaging (MRI), Echocardiogram (echo), and Computed Tomography (CT), but these methods are expensive and not affordable for low and middle-income countries. Therefore, it is important to have an accurate and inexpensive method for diagnosing respiratory illnesses early. An electronic stethoscope can be the solution to reduce costs and improve accuracy Constructing an electronic stethoscope involves digital design, including the SPI interface, register map, synthesis, and post-layout simulations. Analog design consists of 8-bit ADC design and layout, Low Dropout Regulators design and layout, and Bandgap Reference voltage design and layout. This thesis describes the design strategy and the pros and cons of designing analog circuits.
Şebeke bağlantılı çok seviyeli inverterler için yeni bir modülasyon tekniğinin geliştirilmesi ve uygulanması
İki seviyeli dönüştürücüler, kolay kontrol edilebildikleri için güç elektroniği devrelerinde karşılaşılan en temel ve genellikle ilk tercih edilen dönüştürücü tipidir. Bu tip dönüştürücülerin oluşturduğu toplam harmonik bozulma (THD) değerini azaltmak için anahtarlama frekansını arttırmak gerekir. Bu durumda sistem verimi azalmaktadır. Şebeke bağlantılı bir sistemin THD değerini arttırmadan, verimini arttırmanın yolu, çok seviyeli dönüştürücülere geçiş yapmakla mümkün olmaktadır. Çok seviyeli dönüştürücülerin sağladığı bu avantaja karşılık kontrol edilmesi gereken güç elektroniği elemanlarının sayılarının fazlalaşması sistem karmaşıklığını arttırmaktadır. Buna ek olarak, nötr nokta dengesinin dikkate alınmaması durumunda, sistem güvenilirliğinin ortadan kalkması kaçınılmaz olmaktadır. Bu tez çalışmasında, geleneksel iki seviyeli uzay vektör modülasyonundan yararlanılarak, iki veya daha fazla seviyeli dönüştürücüler için genelleştirilmiş yeni bir uzay vektör modülasyonu hesap algoritmasının matematiksel yöntemi ortaya konulmuştur. Bu yöntemde; sistem kaynaklarının arttırılması gerekmemiş, çok seviyeli dönüştürücüye geçiş için sistem tasarımı yalınlaştırılması, iki seviyeli dönüştürücü tasarımı seviyesine çekilmesi ile sağlanmıştır. Yaygın olarak kullanılan iki seviyeli topolojilerin yerine, kontrol edilebilirliği daha karmaşık olmasından dolayı henüz tam olarak yaygınlaşmamış, üç veya daha fazla seviyeli dönüştürücülerin, uygulaması yalınlaştırılmış uzay vektör modülasyonu ile kolay kontrol edilebilirliğinin gösterilmesi amacıyla şebeke bağlantılı üç seviyeli dönüştürücü tasarımı ve uygulaması gerçeklenmiştir. Üç seviyeli dönüştürücü uygulamasından önce iki seviyeli dönüştürücü tasarımı yapılmış, böylece gerçeklenen üç seviyeli dönüştürücü ile aralarındaki donanımsal ve yazılımsal farklılıklar ortaya konmuştur. Donanımsal karşılaştırma yapıldığında; üç seviyeli dönüştürücü tasarımı için kullanılacak tasarım kaynaklarının, iki seviyeli dönüştürücü tasarımı için kullanılacak tasarım kaynaklarından fazla olmadığı görülmektedir. İki seviyeli dönüştürücü için geliştirilmiş yazılım bir anahtarlama periyodu (125 μs) içinde 88-90 μs kadar zaman harcamaktayken, üç seviyeli dönüştürücü için geliştirilmiş yazılım ise bir anahtarlama periyodu (125 μs) içinde 98-100 μs kadar zaman harcamaktadır. Görülmektedir ki, bu tez çalışmasında önerilmiş olan iki veya daha fazla seviyeli dönüştürücüler için genelleştirilmiş uzay vektör modülasyonu hesap algoritmasının uygulandığı yazılımın bir anahtarlama periyodu boyunca harcadığı süre, temel uzay vektör modülasyonu kullanılan iki seviyeli dönüştürücü kontrol yazılımı için harcanan süreden yalnızca 10 μs fazladır. Bu fazla gelen sürenin içine nötr nokta gerilimi dengesi için kullanılan algoritmanın harcadığı süre dahildir.
Human factors and usability of knowledge management for effective e-government
ABSTRACT The concept of e-government, broadly defined as the supporting of public services and information by information and communication technologies (ICTs), is a widely researched topic in the field of human factor usage. This study explains the dynamics behind the effective e-government and the picture of Turkey through digital divide perspective via the sample of the successful e-municipal in Turkey by conducting in-depth interviews with 1 80 people who lives in Kadıköy. The thesis consists of four main parts. The first part includes a general assessment of the extent of adoption and institutionalization of e-government in Turkey. It aims to analyze the current situation, make basic suggestions as to the transition of Turkey to a sound system of e-government and determine primary objectives. This section also stresses the position of Turkey in the digitalized world. Second part of the thesis deals with Human Factors analyzing the influential, cultural and personal approaches, behavioral patterns, possible obstacles that may arise in that regard and the effects of the digital divide. The third part is on the usability of an effective e-government practice. In this part, the primary users of such an introduced system, namely the institutions applying this system and the users who are benefiting will be discussed. The fourth part comprises a Case Study where the application of e-Kadıköy, which is a successful example of e-municipal, is examined in detail. In the survey that is carried out as detailed as possible and where views of the individuals were considered, the results of the interviews and the messages contained therein are investigated thoroughly. VIII
Design of a planner ku band receive antenna array for mobile platforms
A wideband, planar and low profile antenna array is proposed and designed for Ku band direct broadcast from satellite reception for mobile systems. Contrary to earlier designs, the array antenna covers entire downlink Ku-band frequencies (10.8 - 12.75 GHz) for simultaneous reception from multiple transponders. In this thesis, we first propose figure-of-merits (FOMs) to classify antennas in terms of their performances. Among possible antenna configurations, we show that aperture coupled antennas exhibit superior performance when FOM based comparisons are made. Using this single antenna element, we form small arrays first, then large arrays. Besides single antenna element, feed network design also requires a number of challenges to meet low-loss, wideband and phasing requirements. The most straightforward design of feed network is the use of low-loss microstrip lines with power dividers/combiners. However, at Ku band these microstrip lines become too lossy so that array antenna becomes almost useless. A combination of microstrip feed network and waveguide feed network, therefore, is ideal. Then, the main question is how one can form the subarrays and their accompanying waveguide feed network. This thesis answers that question by formulating the subarray and array feed network loss as an optimization problem with constraints on the size and the weight of the array. With all these improvements in the antenna element and the feed network design, the array antenna exhibits 16.5% bandwidth and has 28.4 - 31.3 dBi gain for both vertical and horizontal polarizations. Overall height of the planar antenna with waveguide feed network is under 25 mm. The antenna is built and tested for reception quality. The carrier-to-noise ratio of 9.5 dB is observed for the target transponder (42° E) at 11.9 GHz using IF loop-out of the receiver.
Kızılötesi ışık ile invaziv olmayan kan şekeriölçüm sisteminin geliştirilmesi
Bu tez çalışmasında, kan şekeri seviyesini ölçmek için alternatif olarak invaziv olmayan bir yöntem elde etmek için yakın kızılötesi (NIR) tabanlı bir LED sensör devresi geliştirilmiştir. Sistem 940 nm dalga boyundaki NIR LED'leri kullandı. Analog sinyal, bir filtre ve bir amplifikatör devresi aracılığıyla iletildi. Daha sonra çıkış voltajının ölçüldüğü dijital bir sinyale dönüştürülmüş ve bir Arduino UNO kullanılarak glikoz seviyesi hesaplanmıştır. İn vitro testte, glikoz çözeltisinin değişen konsantrasyonları ile çıkış voltajının değiştiği ve iki değişken arasında iyi bir korelasyon olduğu gözlendi ve bu, in vivo testi motive etti. Deneyimizi üç senaryoya göre gerçekleştirdik: (i) Belli bir miktar glikoz suda çözülür ve ekipmanımızla glikoz konsantrasyonu ölçülür, (ii) katılımcılardan 5 cm3 kan alınır ve kan şekeri seviyesi ölçülür. Cihazımız tarafından bir test tüpünde ve daha sonra bu numune altın standart ekipmana taşınır ve kan şekeri seviyesinin gerçek değeri ölçülür. Tasarladığımız optik ekipmanımızda ölçülen voltaja karşı kan şekeri seviyesi, aralarında oldukça doğrusal bir ilişki olduğunu gösterir. (iii) kan şekeri seviyesi parmak ucu üzerinden ölçülür ve altın standart değer ile karşılaştırılır. Bu durumda ölçüm hatası %25'i geçmez ve kan şekeri düzeyine göre voltaj değişimi doğrusal olmayan bir davranış gösterir. Tasarlanan sistemin maliyetinin düşük olması ve kabul edilebilir bir hata sağlaması, diyabetik hastalar için ticarileştirilmeye uygun hale getirmektedir.
Parçacık sürü optimizasyonu (PSO) kullanılarak elektrik soliton üreteç tasarımı, analizi ve gerçeklemesi
Solitonlar, genliğe bağlı olarak hızın doğru orantılı olarak değiştiği, çarpıştıklarında birbirlerini etkilemeyen, doğrusal olmayan dalga sınıflarıdır. Hızlarının genliğe bağlı olarak değişmesi ve birbirini etkilememesi bu tip dalgaların farklı çalışmalar için kullanımını cazip hale getirmiştir. Askeri radar sistemlerinde, yapay zeka robot çalışmalarında, sinyali uzun mesafelere taşımada, mikrodalga frekanslarında sinyal üretiminde kullanılabilmektedir. Bu tezde soliton dalgası üretimi için doğrusal olmayan iletim hatları (NLTLs) kullanılmıştır. NLTL devrelerinin teorik denklemleri oluşturulmuş ve analiz edilmiştir. Analiz işlemleri sonunda soliton modellemesi gerçekleştirilmiş ve NLTL devrelerini tasarlamak için literatürdeki deneme yanılma yöntemleri kullanmak yerine parçacık sürü optimizasyonu (PSO) algoritması kullanılmıştır. Kullanılan bu algoritma ile 4 farklı merkezi frekans değer paremetresi için optimum sonuçlar oluşturulmuştur. Optimizasyon aşamasından belirlenen devre eleman sayıları ve değerleri için NLTL'ler modellenmiştir. Bu modellerin simülasyonları, 10MHz ile 100MHz arasındaki 10 farklı merkezi frekans değeri için tekrarlanmıştır. Alınan simülasyon sonuçlarından sonra, deneyler gerçekleştirilmiştir. Deneyler 3 farklı merkezi frekans değeri için tekrarlanmıştır. Soliton dalga üretiminde; simülasyonu modellenen ve deneyleri gerçekleştirilen tasarımların uyumlu olduğu görülmüştür.
Akciğer bilgisayarlı tomografilerinden görüntü işleme ve derin öğrenme ile COVID-19 tespiti
Yeni koronavirüs hastalığı (COVID-19) SARS-CoV-2 virüsünün sebep olduğu bir salgın hastalıktır. Bulaşıcılığının çok yüksek olması sebebiyle hastalığın kontrol altına alınabilmesi için enfekte olmuş bireylerin hızlıca tespit edilip izole edilmesi gerekmektedir. Hastalığın tespit edilmesinde sıklıkla tercih edilen yöntem ters transkripsiyon-polimeraz zincir reaksiyonu (RT-PCR) testleridir. Bu testlerin sonuç verme sürelerinin uzunluğu ve başarım yüzdelerinin hastalığın evrelerine göre farklılık gösterebilmesi önemli dezavantajları olarak öne çıkmaktadır. Hastalığın erken evrelerinde hızlı ve doğru bir şekilde COVID-19 tanısı koyabilmek için tıbbi radyolojik görüntüleme yöntemleri de kullanılmaktadır. Bu yöntemler, hasta ile daha az temas gerektirdiğinden daha az bulaş riskine sahiptir. Özellikle, X-ışını (X-Ray) ve Bilgisayarlı Tomografi (BT) görüntüleri üzerinden derin öğrenme temelli yaklaşımlar ile COVID-19'un tespit edilmesini hedefleyen çalışmalar literatürde yoğun ilgi görmektedir. Bu çalışmada, akciğer BT görüntülerini içeren bir veri kümesi oluşturulmuştur. Bu veri kümesi üzerinde derin öğrenme temelli mimariler kullanılarak COVID-19'un tespiti gerçekleştirilmiştir. Derin öğrenme mimarileri kullanılırken iki farklı izlem takip edilmiştir. Birinci izlemde, Basit-CNN ve VGG16 mimarileri kullanılarak ağ derinliğinin başarım üzerindeki etkisi incelenmiştir. İkinci izlemde ise öğrenme aktarımı yönteminin derin öğrenme mimarilerinden VGG19, MobileNet ve DenseNet'in başarımları üzerindeki etkisi araştırılmıştır. Çalışmalar sonucunda DenseNet201 mimarisinin 0,99 test doğruluğu ile en yüksek başarıma sahip olduğu gözlemlenmiştir.
Güncel metasezgisel algoritmalarla denetleyici optimizasyonları
En uygun duruma getirme olan optimizasyon işlemi, mühendislikte en sık kullanılan adımlardan/süreçlerden birisidir. Herhangi bir mühendislik probleminde, belirli koşullar altında mümkün olan alternatiflerin (çözümlerin) içinde en iyisinin seçimi birçok açıdan (verim, maliyet vb.) son derece önemlidir. Matematiksel olarak bakıldığında bu işlem; belirlenen amaç fonksiyonunu istenen tanım aralığında optimum (en uygun) yapan değeri bulmaktır. Bunun için farklı yöntemler (deterministik, rassal vb.) mevcuttur. Son yıllarda optimizasyon problemlerinin çözümünde metasezgisel algoritmalardan geniş alanda faydalanılmaktadır. Bu doğrultuda da birçok metasezgisel algoritma geliştirilmiş ve geliştirilmeye devam edilmektedir. Evrimsel, doğadan esinlenilen (sürü tabanlı, fizik/kimya tabanlı, biyolojik tabanlı, insan tabanlı, vb.) gibi türleri olan bu algoritmalar özellikle büyük ve çok boyutlu optimizasyon problemlerinin çözümünde son derece başarılı sonuçlar vermektedirler. Elektrik-Elektronik Mühendisliği alanında denetleyiciler ve bunların tasarımları çok önemlidir. Temel PID denetleyicilerde ilgili katsayıları ayarlamak için klasik yöntemler (Ziegler-Nichols, Cohen-Coon, vb.) mevcuttur. Gerçekleştirilen tez çalışmasında ise PID türü denetleyicilerdeki katsayıların ayarlanması, güncel altı tane metasezgisel algoritma ile gerçekleştirilmiştir. Bu metasezgisel algoritmalar kullanılarak farklı sistemler için P, PI ve PID denetleyici tasarımları, değişik kriterler (mutlak hatanın toplamı, hata karelerinin toplamı, zaman ağırlıklı mutlak hatanın toplamı, zaman ağırlıklı hata karelerinin toplamı ve özel hata fonksiyonu) altında yapılarak performansları karşılaştırmalı olarak değerlendirilmiştir.
Dış mekân elektromanyetik maruziyet seviyesinin yapay sinir ağı ve bulanık mantık ile tahmini
Elektromanyetik (EM) kaynakların insan hayatındaki önemi teknolojinin gelişmesiyle birlikte artmaktadır. Elektromanyetik radyasyon çevrede var olan elektronik cihazları ve daha da önemlisi insan sağlığını olumsuz yönde etkileyebilmektedir. Günümüzde bahsedilen sorunları tetikleyebilen yapılardan biri olan baz istasyonu bir bölgedeki insan yoğunluğu ile orantılı olarak radyasyon yaymaktadır. Hastaneler, konser alanları, askeri kışlalar, okullar, alışveriş merkezleri gibi insanların yoğun olarak bulunduğu alanlarda elektromanyetik maruziyet diğer bölgelere göre daha fazladır. EM seviyelerini izin verilen sınırların altında tutmak için bu alanlardaki elektromanyetik radyasyon yoğunluğu hakkında sürekli endişe göstermek önemlidir. Bursa Uludağ Üniversitesi Görükle Yerleşkesi'nde 55000'den fazla öğrenci ve her gün binlerce kişinin ziyaret ettiği bir tıp fakültesi hastanesi bulunmaktadır. Bu nedenle EM düzeyini tespit etmek ve bu değeri ulusal standartların altında tutmak önem arz etmektedir. Bu çalışmada, kampüs genelinde EM seviyelerini incelemek için insan yoğunluğunun çok yüksek olduğu noktalarda elektrik alan şiddetleri statik olarak ölçülmüştür. Elde edilen ölçümler sonucunda yapay sinir ağı ve bulanık mantık yöntemleri ile elektromanyetik alan şiddeti tahmini için modeller geliştirilmiştir. Oluşturulan modeller kullanılarak ölçüm yapılmayan noktalarda elektromanyetik alan seviyesi tahmini yapılmış, ölçüm sonuçlarıyla doğrulanmış ve modellerin performansları karşılaştırılmıştır Ölçüm sonuçları ve tahmin sonuçları, ulusal (BTK) ve uluslararası (ICNIRP) standartların belirlediği sınırlar içinde değerlendirilmiştir. Son olarak, ölçüm tahmin sonuçları, ölçüm noktalarına ve ortalama elektrik alan şiddeti yoğunluğuna göre haritalandırılmıştır.
Radar kesit alanının azaltılması için ultra geniş bantlı frekans seçici yüzey tasarımı
Frekans seçici yüzeyler (FSY), belirlenen frekans aralığında yüzeye gelen elektromanyetik dalgaların iletim, yansıma veya soğurma özelliği göstermesini sağlayan periyodik yapılardır. Bu çalışmada birden fazla frekans bandını kapsayacak şekilde (C-bandı (4-8 GHz) ve Xbandı (8-12 GHz)), ultra geniş bant aralığına sahip, halka ve kesikli artı şekillerinin birleştirilmesiyle elde edilen yeni UGB FSY ile radar kesit alanının (RKA) azaltımı amaçlanmıştır. Çalışmanın ilk kısımlarında halka ve kesikli artı şekilli yapılar tek tek analiz edilmiştir. Analiz sonucunda en verimli olan iki yapı birleştirilerek elde edilen yeni FSY ile -10 dB kazanç değerinde bant genişliği arttırılmıştır. Önerilen yapı üzerinde temel parametreleri değiştirerek, ultra geniş bantlı (UGB) radar sistemlerinde kullanılabilecek iyi zayıflama düzeyine sahip, dalganın geliş açısından bağımsız ve bant durduran filtre karakteristiğine sahip yapı elde edilmiştir.
Mikrodalga frekansında elektromanyetik soğurucu tasarımı ve optimizasyonu
EM soğurucular, günümüz askeri ve sivil endüstrilerinde çok sayıda uygulama alanı bulmaktadır. Belirli bir EM spektrum bandı içinde geniş bir frekans aralığında iyi yutuculuk özellikleri ve ince katman yapısına sahip mikrodalga soğuruculara duyulan ihtiyaç da bununla birlikte artmaktadır. Bu çalışmada, Kel Kartal Arama Optimizasyonu Algoritması kullanılarak çok katmanlı yapılar için EM soğurucu tasarımı gerçekleştirilmiştir. Seçilen frekans aralıklarında farklı geliş açıları ve polarizasyonlar için literatürde yer alan İyileştirilmiş Parçacık Sürü Optimizasyonu, DE, CFO vb. ile karşılaştırmalar yapılmıştır. Farklı malzeme veri setleri kullanılarak çeşitli tasarımlar ile en uygun tasarımın bulunması hedeflenmiştir. Kel Kartal Arama Optimizasyonu ile 19 -37 GHz frekans bandında, çok katmanlı mikrodalga soğurucu için en uygun tasarımların elde edilmesi amaçlanmıştır. Bu optimizasyonların kullanım amacı, önceden tanımlanmış olan bir veri tabanından uygun malzeme katmanlarını seçerek soğurucunun istenen frekans aralığında en düşük maksimum yansıma katsayısına sahip ve en ince kalınlıkta soğurucu tasarımını gerçekleştirmektir. Optimizasyon sonucunda alınan her bir varyasyon için sayısal en iyi sonuçlar sunulmuştur. Bu tez çalışması gerçekleştirilirken MATLAB R2021b programından yararlanılmıştır.