Prof. Dr. Arif Nacaroğlu danışmanlığındaki tezler
15 tez · Gaziantep University
A methodology for access control to ibeacon network
By developments in the technology, new products that ease users' work are developed. iBeacon is a recently developed by Apple, makes major changes in our lives especially in marketing and selling areas. It allows mobile applications to be aware of users' locations by using Bluetooth Low Energy (BLE). iBeacon devices send one way communication signals to mobile devices for advertisement messages etc. It brings many advantages to users who use the application, but it has many obstacles as well. Security of the iBeacon devices has crucial importance since the use of same physical machine by multiple users may result in bad consequences. Authentication of iBeacon needs to be taken under consideration since without secure authentication, malicious users may change the signals or the messages sent by iBeacon devices. In this thesis, we focus on the challenges during the authentication stage of iBeacon, by proposing an authentication model which prevents existence spoofing attacks while iBeacons send signals to mobile devices.
Design of the fast Gaussian filters for image processing
Image processing is a method used for execution of some operations on image transfered into the difital medium by transducer. Digital image may be processed in order to increase the quality of the picture as well as to get some information from the Picture such as identification. But the application area of the image processing is not limited only for enhancement. It can be used in many different areas like engineering applications, medical measurements, digitization of the information, space studies, natural sources surveys taking the satellite images, security, film industry, military applications like target detection, otonomous aerial vehicles, night vision, etc. In this thesis, Fast Gaussian filter is studied for its application to image processing. The coefficients of the Gaussian filter are changed by a specific algorithm, the changes in the images are observed for image processing applications. The attension of this study is focused on to obtain the most suitable coefficient for each image.
A new solvable class linear time varying system
In this thesis, a new definition of classification has been done, with regard to the time- varying linear systems can be solved and the solution method for the new proposed class has been described. There isn't a general analytical method of solution in time-varying linear systems and being solvable of these systems is related to their being able to turn into time-invariant systems, applying certain transformations. In the studies conducted to date, in general, it has been shown that two groups of the time-varying linear systems are solvable, in addition to this, some special systems that can be transformed into these groups, although not belonging to one of the these to groups, have been introduced. Determination of the system's being solvable is directly related to the structure of the system's eigenvalue and also, in this study, on the basis of eigenvalues, a new solvable group has been identified.
The fluctuations of the switching instants in the switched networks
In this thesis, two studies on the fluctuation of the switching instants in the switched networks are presented. The first study is on the effect of the deviation of the switching instants on the transfer characteristic function of the switched networks. And in the second one, a statistical approach is proposed for the analysis of the noise resulted from the random fluctuation of the switching instants in switched networks. Since the switched capacitor networks (SCN's) are the most popular and used structure among the switched circuits, the analysis of the effects of the switching instant fluctuations is carried out for these networks throughout this thesis. But the obtained results can also be easily applied to the other switched circuits. In the first study, by using the phase transfer matrix entries, the system transfer function including the switching instants is given. Two programs have been developed to find the phase transfer matrix entries and the system transfer function. By executing the program for various switching instants, it can be seen that the transfer characteristic of a switched capacitor network is highly affected from the fluctuation of the switching instants. Moreover, from the simulation results it is seen that a switched capacitor integrator can be used as a phase filter by only changing the switching instants of the clock signal applied to itself. IVAs for the second study, by choosing the duration between the switching instants as the random variable of the statistical process, the conventional Gaussian distribution is first modified. Then, an analytic expression for the calculation of the output noise, based on the modified Gaussian distribution function, presented in SC networks due to the random fluctuation of the switching instants is obtained. The found formula can also be applied to any switched network. The simulation result makes the theoretical results meaningful.Key words: Switched capacitor networks, Random fluctuation, Noise analysis, Modified Normal distribution.
Software coincidence of general purpose data acquisition cards with Matlab toolbox
Data acquisition systems are generally used to acquire physical data by usingtheir concerning software and hardware. There are many components of professionalmeasuring systems such as data acquisition hardware, transducers, signalconditioners, computer and software. In some professional testing and measurementsoftware packages are already installed on hardware components of data acquisitionsystems to simplify the applications. However, these applications are used for smallsize data acquisitions, for very low voltages and frequencies.In this thesis it is aimed that managing software of data acquisition cards andMatlab Data Acquisition Toolbox on one platform (Data Acquisition Control Panel).The preparing step of hardware initialization is made by written interface programautomatically. All requested DLL (Dynamic Link Library) for Matlab can beregistered, modified or automatically created, if it is not present on directory.Creating analog inputs and their necessary steps are reduced and linked to specialbuttons on Data Acquisition Control Manager. All unnecessary log filesautomatically removes from hard drives when digital oscilloscope program ends. It isnot necessary to be familiar Matlab, Data acquisition Toolbox and hardware interfaceprograms to use digital oscilloscope program of Matlab.Data Acquisition Control Manager is linked to softscope (Digital OscilloscopeProgram) to complete easy measurements and testing.Keywords: Data Acquisition, Test and Measurement, Matlab, Data AcquisitionToolbox
Enhancement of speech communication between human and machine
Speech is a most natural and efficient way to exchange information for humanbeings. To create this information exchange between machine and humanbeing and to make a real ?intelligent computer,? it is important that themachine can ?hear?, ?understand,? and ?act upon? spoken information, andalso ?speak? to complete the information exchange. Therefore, ?speechrecognition? is essential for a computer to reach the goal of natural human-computer communication. Speech Recognition is a fascinating field spanningseveral areas of computer science and mathematics. Reliable speechrecognition is a hard problem, requiring a combination of many techniques;however modern methods have been able to achieve an impressive degree ofaccuracy but the improved algorithms still have some problems.This project attempts to examine one of those techniques, and to apply them tobuild a sample voice recognition system for Turkish. The goal of this thesis isto be able to recognize isolated words, spoken individually. In this work, thespeech data is loaded in to the computer using data acquisition card. This datais processed by software prepared in MATLAB. Linear Predictive Coding andDynamic Time Warping method is used in the program algorithm.Keywords: signal processing, speech recognition
Investigation of a new method for direct IIR digital filter realization using flattened coefficients
A new direct method for the conversion of FIR transfer function to IIR transfer function using flattened coefficients is presented. The non-linear equations are obtained by the long division coefficients of the IIR transfer function. The given coefficients of the FIR filter are equated to these equations and solved by using the non-linear system solution algorithms. To reduce the error between the IIR filter and given FIR filter, an error minimization algorithm is used and the presented method is compared with the other approximation methods. The effects of binary flattened coefficients are investigated. Keywords: FIR filter, IIR filter, long division
Design of high-quality low-order nonrecursive digital filters using the window functions
The aim of the thesis study is to present a new window function (or simply window) to provide better nonrecursive filter characteristics than the filters designed by the windows in the literature. To achieve this aim, four windows - namely Exponential, Cosh, modified Cosh and modified Kaiser - are proposed.The Exponential and Cosh windows are derived by replacing the modified Bessel function of first kind of order zero in the Kaiser window with the exponential and cosine hyperbolic functions, respectively. Unlike the Kaiser window, they have an advantage that they have no power series expansion in their time domain representations. The spectrum design equations are established for these windows. The simulation results show that compared to the Kaiser window for the same window length and mainlobe width, they provide very good sidelobe roll-off ratio characteristic - which may be useful for some applications. But, in terms of the ripple ratio, the Kaiser window performs better results. As for the comparison with the three-parameter Ultraspherical window, the Exponential window performs better in ripple ratio characteristic for narrower mainlobe width and larger sidelobe roll-off ratio with fixing the window length, mainlobe width and sidelobe roll-off ratio. For wider mainlobe width and smaller sidelobe roll-off ratio, it performs worse ripple ratio. The Cosh window performs better ripple ratio characteristic for wider mainlobe width and larger sidelobe roll-off ratio. But, for narrower mainlobe width and smaller sidelobe roll-off ratio, the Ultraspherical window provides better results.Third and fourth proposed window functions are the modified Cosh and modified Kaiser windows which are derived by proposing a third adjustable parameter to the two-parameter Cosh and Kaiser windows, respectively. The simulation results show that they provide superior ripple ratio characteristics compared to the Kaiser and Cosh windows.As for the applications of the proposed windows in the design of nonrecursive digital filters, the filter design equations for the Exponential and Cosh windows are obtained, and then they are compared with the Kaiser window. The simulation results show that the Kaiser window provides better results in terms of the minimum stopband attenuation. But, the Exponential and Cosh windows are better in terms of the maximum stopband attenuation which may be useful for some applications. The optimum solutions for the modified Cosh and modified Kaiser windows are found, and then the simulation results show that these two modified windows performs very good results in terms of the minimum stopband attenuation compared to the Kaiser window.The comparisons with the well known window functions in literature show that the proposed modified Cosh and modified Kaiser windows can provide the best quality in terms of the minimum stopband attenuation for a fixed filter order, and also provide lowest filter order for a fixed minimum stopband attenuation.
Aktif ve pasif filtrelerin bilgisayar destekli dizaynı ve uygulamaları
Bu tezde, yaklaşım metotları ile istenilen filtrelerin gerçekleşmesini sağlayan bilgisayar programı sunulmuştur. Çalışılan bilgisayar programı, istenilen filtrenin gerçekleşmesini elde etmek için, Butterworth, Chebyshev ve Ters Chebyshev türü yaklaşım metodlarını içermektedir. Eliptik tür yaklaşım metodu ile alçak ve yüksek geçiren filtrelerin transfer fonksiyonu elde edilecektir.Filtre devresini gerçekleştirecek devre elemanlarına ulaşmak için 3 tip yaklaşım metodu kullanılmıştır. Bunlar Butterworth, Chebyshev ve Ters Chebyshev yaklaşımlarıdır. Eliptik yaklaşım kullanılarak transfer fonksiyonu elde edilmiştir. Özel filtreleri karakterize eden: geçiren frekans (wp), geçme bandı dalgalanma değeri (Kp), durduran frekans (ws), durdurma bandı dalgalanma değeri (Ks) ve üzerinde çalışılacak yaklaşım metodu, kullanıcı tarafından programa girdi olarak verilecektir. Visual C# ile yazılan program, özel filtreyi gerçekleştiren devre eleman değerlerini hesaplayarak, ekrana devre ile beraber verecektir. Yaklaşım metotlarında, devre eleman değerlerini vermeyi sağlayan matematiksel fonksiyonların sonuçları, ekranda görülebilecektir.Anahtar kelimeler: geçiren frekans (wp), geçme bandı dalgalanma değeri (Kp), durduran frekans (ws), durdurma bandı dalgalanma değeri (Ks).
Design and simulation of distance protection relay for high voltage transmission line using artificial neural networks
This thesis presents an approach to design a distance protection relay that performs the function of detection, classification, and localization of the fault on high voltage transmission line. The method is based on combination of discrete wavelet transform as a technique for processing signal and artificial neural networks using the extracted features from current signals of fault by discrete wavelet transform as inputs to artificial neural networks that will be trained by using back-propagation algorithm to differentiate between non faulty and faulty cases in addition to, classification of fault type, and localization of the fault for power transmission line. Matlab/Simulink is utilized for designing the model of power transmission line and simulating the different types of faults for various fault resistance and distance as well as design, training, and testing of neural networks for detector, classifier, and locator.
Minimization of the finite word length noise in the iir digital filters using truncation and roundation
In this thesis, effects of finite word length on Infinite Impulse Response digital filters (IIR Digital Filters) are studied and it is tried to approach ideal filter characteristics with the aim of using minimum bit number(s) and getting minimum error. The coefficients of the filter are changed by using truncation and/or roundation approximation to obtain the acceptable gain characteristic with shorter binary coefficients.
A new window function for fir filter design
FIR (Finite Impulse Response) filter is an important research area in digital signal processing and filtering. FIR filters provide easily design at linear phase, which means delay the input signal but is delayed without distorting its phase. In addition to these advantages, FIR filters don?t need to feedback. At FIR filters, to design a filter means to select the coefficients such that the system has specific characteristics. Generally, filter specifications refer to the frequency response of the filter. There are different methods to find the coefficients from frequency specifications. One of the important method is the series expansion and windowing. The window method for digital filter design is fast, convenient, and robust. In this study, a new window function proposed to calculate the FIR filter coefficients obtained by using Fourier Series Method. In basic, proposed window is a modified version of Gaussian window function and Hamming window function. Combination of these windows has resulted with more parameters which means the controlling of window function is more flexible. Keywords: Window functions, Gaussian window, Hamming window, Modified Gaussian window, Modified Hamming window, Finite Impulse Response (FIR) Filter design.
Development of a solvable class of linear time varying systems
There have been various solution methods including Laplace transform developed for Time Invariant Linear Systems (TILS). However, there isn't a general analytical solution method for Time Varying Linear Systems (TVLS). To solve these systems, we should find suitable transfer matrices to transform TVLS into TILS so that analytical solution becomes possible. In this study, we have added a new class to knownsolution classes (A1 Class,Ah Class,HG Class) to transform TVLS into TILS and the proposed solution class is presented.The solvability conditions for the linear system is reviewed and the relation between solvability conditions and the eigenvalues are given. Transformation method of the new class of linear systems is derived and supported by some examples. Keywords: Time varying linear system, Solvable linear system
Design of digital iir filter with finite word length coefficients using genetic algorithms
In this work, a method for the application of the genetic algorithm to design a digital infinite impulse response (IIR) filter with finite word length binary coefficients is given. Program selects the best coefficients and minimum order among the coefficients and orders in the population. We have prepared a program that allows selection of the most suitable solution from within the pool of possible solutions that are created from normalized and finite word length coefficients. This program provides the most suitable coefficients and optimal order is selected. Local search algorithm which aims to reducing the number of iteration of genetic algorithm have been proposed. Coefficients have been selected in the pool that are occurred less than 6 bits in size and consist of sum of power of 2 coefficients. So in practice, aimed to increase the opportunities to use and quick response of the filter.
A new and efficient window function for digital FIR filter design
In this thesis, a new window function for digital FIR filter in addition to studied in literature is presented. Some of the well-known and widely used window functions given in the literatures are Kaiser window, Cosine hyperbolic window, Exponential window, Saramaki window, Ultraspherical window, etc. Each window function has advantages and disadvantages compared to their specific application areas. Digital filters are commonly used for application of digital signal processing. The linear phase characteristics of the finite impulse response (FIR) filters make them preferable to infinite impulse response (IIR) filters although their lengths are more. The stability property of FIR filters is the another advantage. There are some methods of designing the FIR filters starting with the given magnitude and/or phase specifications. The one of efficient method for FIR design is to use the Fourier Series. Since the Fourier Series representation of the specified characteristics are infinite and the FIR filters have finite length, to prevent the Gibbs' Oscillations, the window function are used. Briefly, combination of modified Cosh window and Blackman window is proposed in this thesis. Especially, sharper transient and lower magnitude of attenuation are obtained. To find best optimized values, most of mathematical calculations are analyzed, then some specific parameter values are found for proposed window function. Key Words: Finite impulse response (FIR) digital filters, window methods, Blackman window, modified Cosh window.