Ph.D. Theses

Permanent URI for this collection

https://digitalarchive.library.bogazici.edu.tr/handle/123456789/11648

Browse

Now showing 1 - 20 of 95

Stochastic adaptive receding horizon controllers
(Thesis (Ph.D)- Bogazici University. Institute for Graduate Studies in Science and Engineering, 1982., 1982.) Yaz, Engin.; Istefanopulos, Yorgo.
In this thesis, the deterministic, stochastic and stochastic adaptive coritrolpossibilities based on the method of receding horizon is examiried. The receding horizon method assumes a fixed horizon length for feedback law calculation at each step. Therefore, the feedback law is optimal in one-step-ahead manner and the feedback gain is constant. The other advantages are of not having to choose the state penalization matrix and of replacing the solution of Riccati equation by a linear one. We alleviated some problems associated with the practical use of this method, such as calculation time and singular state transition matrices by some fast algorithms and non-zero set points by modification of the basic equations. Modelling the system in state space innovations representation or transforming it to this form if it is not modelled in innovations form originally, solves the problem of state reconstruction under noise effects. The overall design enjoys the separation property, that is, of having a separate design for control and estimation parts. In the case of some unknozn parameters in the system equations, our controller works using the state estimates, found by utilizing the parameter estimates, in the control law, and parameter estimates, found by using the state estimates, in the feedback gain calculation. This controller with this enforced certainty equivalence property enjoys many favorable characteristics such as refraining from the use of Riccati equation in control, matrix update equations for state and parameter estimation uncertainties, external perturbation signals to secure stability, and trial and error procedures in the choice of state penalization matrices. Moreover, the method is general enough to control with any prescribed control strength, multi-input, multi-output systems under noise effects, modelled in difference equation from with multi-parameter uncertainty.
Modern spectrum estimation
(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1983., 1983.) Aktar, Mustafa.; Istefanopulos, Yorgo.
Sensitivity theory application to the numerical solutions of the general optimal control problem
(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1983., 1983.) Eksin, İbrahim.; Istefanopulos, Yorgo.; Eyler, M. Akif.
This study proposes various efficient numerical methods for the general optimal control problem. The basic feature of the methods developed here is that they somehow exploit the ideas and the concepts of the sensitivity theory. First, a new method for solving TPBVP, which is met in seeking an open-loop solution for optimal control problems, is developed. This method can be briefly expressed as an iterative procedure which is based on trajectory sensitivities with respect to initial conditions. In the second part of the study, various numerical methods are developed for the closed-lobp solutions of general optimal control problems using performance index sensitivity functions with respect to controller parameters. These new methods may be treated in two categories : 1) Apriori polynomial approximation methods; Here the basic assumption is that controller parameter function is assumed to be formed by a polynomial function. 2) Aposteriori polynomial approximation method; In this method a sequence of subproblems are created using some intrinsic properties of the previous method. The values of the results of the subproblems are then used in the formation of the optimum controller parameter function.
Multiprocessor optimizations: |interconnection and task assignment
(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1985., 1985.) Erdim, Füsun.; Istefanopulos, Yorgo.
Effective spreading of the use of muLtiprocessors, -or distributed processing in general-, and achieving the potential advantages of this new design option require various hardware and software-related probLems to be solved. This study is a research on two basic probLem areas, namely the Interconnection and the Task Assignment in Multiprocessors. Any multiprocessor system that employs more than one processor for a singLe job must be designed to allow efficient communication between processors, so that the advantages of multiprocessing is not negated by inefficient communication. As the number of processors grows, the interconnection design becomes more cruciaL as crossbar or fully-connected schemes become impractical. Thus, from a realizability point of view a partially-connected structure is desirable, which, however, in turn, introduces the problem of variable interprocessor distances, complicating the task assignment process. In the first part of this study, PON (Processor Omega Network), a partially- connected, multistage processor network with desirabLe implementation and communication properties is proposed and evaluated. In any distributed processing environment, except for identical processors forming a fully-connected network of uniform interprocessor distances, optimal assignment of software modules comprising a task to processors of the network is essential for minimum-time completion of the task and this can be achieved by balancing two conflicting factors; minimization of interprocessor communication and maximization of load balance of processors. In addition to the complexities of the previously studied resource limited task-assignment environments, partially-connectedness introduces the new interrelated problems of indirect data transfers, availability of intermediate processors, and data routing when more than one path is available between non-adjacent pairs. Two different performance measures are proposed for the two operation environments considered. The minimum port-to-port time (PTP) criterion produces optimal assignments in single-run environments, whereas the optimum performance in a multi-run operation mode is achieved by minimizing the least re-initiation period (LIP), which is equivalent to maximizing the overlap between successive task executions. The characteristics of the objective functions, the number of constraints, and the precedence reLations dictated an aLgorithmic solution to the assignment problem. An analytical model is developed to describe the task assignment environment considered in this study, and based on the model components and the proposed objectives, the optimization problems for both environments are formulated. Some possible methods for storage-and-processing efficient representations of hardware and software are investigated and the task assignment algorithm for partially-connected networks (PCTAA) is presented and the methods and modifications to reduce its computational complexity -related to the structure of networks and tasks- are discussed in order to extend its use to analysis of larger systems.
Novel results on frequency estimation and statistical characterization of three spectral estimation techniques
(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1985., 1985.) Anarın, Emin.; Sankur, Bülent.
The problem of detecting a sinusoidal signal in white noise and estimati on of its parameters is essentially a problem in signal processing such as radar, sonar, biomedical, etc. In this dissertation, various modern spectrum estimation approaches, Maximum Entropy Spectral Analysis (MESA) spectral moments and analytic signal techniques and their statistical characterization have been investigated and formulated in detail. The development of modern spectrum estimation known as parametric techniques for estimating parameters of sinusordal signals in white noise is important. Therefore the parameter estimation technique based on previously appeared and as well as some other newly developed modern spectrum estimation procedures have been presented in this dissertation. Comparative performances and drawbacks of most of the parametric techniques known as Maximum Likelihood (ML), Maximum Entropy (ME), Pisarenko, Kumerason, Prony methods used in frequency estimation have been summarized. The analytic signal model called as Argument method to estimate frequency and bandwidth of a sinusoidal signal is studied in detail. New expressions related to the expected value, variance and probability density function of estimate are derived analytically.
Study of evaporated amorphous silicon films by frequency dependent conductivity and phase shift analysis of modulated photocurrent methods
(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1989., 1989.) Çil, Celal Zaim.; Tansal, Sabih.
A.c. and d.c. conductivity measurements have been performed on amorphous silicon films produced by electron gun evaporation. The a.c. measurements have been performed at frequencies between 100 Hz and 2 MHz, and at temperatures between 150K and 400K. The d. c. behavior of the samples obey the T- 1/4 law between 77K and 250K. The a. c. conductivity of the films are well represented by the form Aw s , where A and s are determined to be temperature dependent parameters. The data are discussed in terms of classical models based on, pair approximation and a unified theory, the extended pair approximation, EPA. Although the a. c. behavior can be approximately explained by the Correlated Barrier Hopping model below 200K, the temperature and the frequency dependence are stronger than any classical model predicts. The a.c. data show a perfect agreement with the quasi-universal law predicted by the EPA calculations. However, quantitative calculations with the EPA results give unreasonable values for both the decay parameter a, and the rate parameter Ro. One of the major problems of the method of Phase shift analysis of modulated photocurrent, PSAMP, for studying the density of states in the energy gap of amorphous semiconductors has been the determination of the energy scale corresponding to this DOS profile. A new way of dealing with this problem is presented. A computer analysis is used to confirm the validity of this method and to demonstrate how it can be used. A simulation that is used to determine the sensitivity of the PSAMP method to the differences in the fine scale structures in the DOS distributions is presented. Four DOS distributions are considered and the expected data are obtained. The results show that the PSAMP method is very sensitive to such fine features in the DOS distributions. A comparison is also made with the sensitivity of other techniques commonly used in the determination of the DOS profiles.
Design of multidimensional perfectreconstruction filter banks with compression applications
(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 1996., 1996.) Celasun, Işıl.; Sankur, Bülent.
The acoustics, vibration, tacho voltage (tacho is a part of the universal motor used to get a feedback from the motor to control the speed), feeding voltage and the supply current signals reflect the health of the motor. When there is a diversion from the normal operating situation these signals change their characteristics. By analyzing these signals and comparing the results with the reference data obtained from the non-defective motors, the anomalous behaviour can be detected. These signals can be used individually or together in the fault detection of the motors. A number of techniques have been developed which monitor certain parameters of the motors, allowing its health to be determined. These monitoring techniques are known as Machine Condition Monitoring. Condition monitoring technique can often be separated into two main categories, those being invasive and those being non-invasive. Invasive monitoring, as the name suggests, involves the disassembly of the motor in question, whereas the non-invasive techniques allow the health of the motor to be obtained while the motor is still in its normal operation. Modem signal processing techniques allow us to see much more deeply into the operations of plants and processes, particularly when we base the prpcedure on a fundamental understanding of the mechanics of operation of the machine. There are var:ious kinds of signal processing tools to analyze the signals obtained from the motors. In this work we aim to detect the noise problems and identify the various faults in the universal motors by monitoring and analyzing the sound and vibration signals. This method contributes a cheap approach to fault identification of the sound and vibration signals coming from the defective motors, that could originate from a defective brush and/or commutator, defective bearings, an unbalanced rotor or from a tom-folio of the rotor) are compared against the sound database of non-defective motors. For fault identification purposes extensive use of periodogram spectra was made. Feature extracted from spectra were assessed as for their discriminatory power. since rotating machinery gives rise to resonant spectra, spectral peaks were found to be the best set of evidence to classify between motor fault types. Preliminary statistical results with spectral peaks indicate that it is feasible to develop a noninvasive tool based on acoustics / vibration data for fault classification.
Development of a high level synthesis tool specialized on fir-based multirate systems
(Thesis (Ph.D.)- Bogazici University. Institute for Graduate Studies in Science and Engineering, 1999., 1999.) Yurdakul, Arda.; Dündar, Günhan,; Tansal, Sabih.
Digital Signal Processing (DSP) is the most studied area in design automation, because it is one of the most well-established branches of electrical engineering for several years. In the last few years, it is stimulated by the progression of multirate techniques. The key property of multirate algorithms is their computational efficiency. In this thesis, a silicon compiler is developed to reduce design time for the hardware realization of FIR-based multirate DSP algorithms. This is a brand new study, because there does not exist a silicon compiler of this type according to our knowledge. Although multirate algorithms contain decimators and interpolators changing the effective sample rate, the design of synchronous systems using a single-clock signal is possible by this newly developed tool. The designer can achieve this by folding nodes of similar type into a single node. Additionally, the FIR filters followed by a decimator or following an interpolator can be entered as a single node while defining a system at the input of the tool. Also multiplications with the tap coefficients in FIR-based nodes in a fold are handled at the same time to exploit common terms so as to realize those multiplications without multipliers. As a result, the tool produces very efficient layouts in terms of area, power and clock signals. It can also determine the quantization levels of tap coefficients in FIR-based nodes and fractional parts of data bus if the system output error is specified. It also handles module selection under given power, area and delay constraints and scheduling like other well-known silicon compilers. The compiler is programmed to process bit-parallel-digit-serial architectures.
Multilayer perceptron neural network in analog VLSI-A system level study
(Thesis (Ph.D.)- Bogazici University. Institute for Graduate Studies in Science and Engineering, 1999., 1999.) Öğrenci, Arif Selçuk.; Dündar, Günhan,; Balkır, Sina.
Analog neural networks exhibit a potential for proper/suitable hardware implementation of artificial neural networks. Their advantages such as small size, low power and high speed, however, are seriously questioned/confronted by the difficulty in the training of analog neural network circuitry. Especially, hardware training by software, i.e., training of the circuitry by software based on models, so as to avoid on-chip and chip in-the-loop training methods, is threatened by circuit nonidealities and variations at outputs of identical blocks. The performance of the analog neural network is severely degraded in the presence of those unwanted effects caused mainly by statistical variations in the production process. We propose a new paradigm for the backpropagation algorithm in hardware training of multilayer perceptron type analog neural networks. The variations at outputs of analog neural network circuitry are modeled based on the transistor level mismatches occurring between identically designed transistors. Those variations can be used as additive noise during the training, and it has been shown that this will increase fault tolerance of the trained neural network drastically. The method has been compared to the method of injecting random noise, and our method outperforms the latter where injecting random noise is seen to be inadequate for establishing a satisfactory level of fault tolerance in the presence of mismatch based variations. The concept of mismatch based variations has been verified by measurements on our test chip.
Variable structure systems theory based training strategies for computationally intelligent systems
(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2000., 2000.) Efe, Mehmet Önder.; Kaynak, Okyay,
Noise rejection, handling the plant-model mismatches and alleviation of structured or unstructured uncertainties constitute prime challenges that are frequently encountered in the practice of systems and control engineering. One way of reducing the adverse effects of the stated difficulties and obtaining a good tracking precision is to utilize the techniques of variable structure systems theory, which offers well formulated solutions particularly to problems containing uncertainty and imprecision.In this thesis, variable structure systems theory based training strategies of computationally intelligent systems are discussed. Two approaches are developed for alleviating the above mentioned difficulties. Additionally, the learning rate selection problem is treated from the point of variable structure control.In the first approach described, a dynamic parameter adaptation law is derived and the applicability of the algorithm is discussed. The analysis presented aims to extract the conditions for establishing equivalence between sliding mode control of the plant and sliding mode learning in the controller. The second method is based on the selection of an extended Lyapunov function, by the use of which the sensitivity of the cost measure to the adjustable parameters are minimized together with the half squared error measure. Lastly the selection of the learning rate for three different gradient based parameter tuning strategies are discussed. The objective of the learning rate selection is to drive the plant to a sliding mode while the output of the controller is driven to a similar regime.The performances of the methods developed are assessed on the dynamic model of a two degrees of freedom direct drive SCARA robotic manipulator, whose dynamic equations are assumed to be unknown throughout the results presented. In the simulations, the alleviation of the adverse effects of observation noise and varying payload conditions are studied.
Reconfiguring sliding-mode controller for underwater vehicles
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2004., 2004.) Demirci, Ufuk.; Özçaldıran, Kadri.; Kerestecioğlu, Feza,
The challenging control problem of underwater vehicles is addressed in this study under excessive sea wave disturbances at shallow submerged operation. The systems are becoming more and more complex with the effect of rapid technological developments. These complex systems require fault tolerant control systems in order to accommodate with the uncertainties caused by faults or environmental changes. In this doctoral dissertation, a novel re-configuring controller is proposed based on sliding mode methodology for fault or disturbance tolerant control. Re-configuring control with sliding mode technique annuls the unwanted effects of system uncertainties caused by faults and environmental effects and avoids the chattering by increasing the robustness of the controller when it is necessary. As a result acceptable performance is obtained in case of system uncertainties and the chattering phenomenon is eliminated for the nominal system situation.
Optimal nonlinear controller design for flexible robot manipulators
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2006., 2006.) Doğan, Mustafa.; Istefanopulos, Yorgo.
Developing nonlinear adaptive and robust controllers for a two-link flexible robot arm is the main objective of this research. Two different modelling techniques are used to overcome model accuracy problems and these two are evaluated comparatively. These are FEM(Finite Element Method) as a reduced order approximate model and PDE (Partial Differential Equations) approach as an exact model. Since FEM model needs modal truncation, unknown disturbances can excite neglected high-frequency modes generated by the nonlinearities of the plant. The second approach is modelling the flexible robot arms by PDE which are known to provide exactness. In order to improve the important features of flexible links such as low mass and moments of inertia and high natural frequencies, optimal shape design can result in nonuniform cross-section of links. Furthermore, a high fundamental frequency is desired since it implies a large bandwidth that will allow for fast motion without causing serious vibration problems and for stable endpoint control. The main results of the study are robust regulation of the rigid modes and suppression of elastic vibrations of the flexible robot arm. The dynamic state feedback controller is used to achieve this goal in FEM approach. In the first part of this research the adaptive internal model approach, in parallel with a robust stabilizer, has been modified to manage totally unknown disturbances that can include neglected higher modes of the uniform flexible links as well as large parameter uncertainties such as tip mass changes. The stabilizer part of the controller which has been introduced for the first time in this research in conjunction with nonlinear systems, is optimized successfully with a new efficient evolutionary algorithm. In the second approach (PDE) of this research, the control of a two-link flexible arm with nonuniform cross-section by design is improved by employing the Lyapunov method. LaSalle's invariance principle, extended for infinite dimension, is used in order to prove the asymptotic stability of the closed loop system without any modal truncation as opposed to former approaches in literature. Besides, large parameter uncertainties such as tip and hub mass changes in PDE approach are also handled effectively by the proposed nonlinear controller.
Solutions to restrictions of the current conveyor based circuits
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2006., 2006.) Yüce, Erkan.; Çiçekoğlu, Oğuzhan.
Current conveyor (CC) is one of the most widely used versatile active components. It potentially provides higher accuracy, wider bandwidth, greater linearity and larger dynamic range. The advantages of the CCs have brought widespread usage in some technical areas such as current-mode analog communication systems and signal processing. It was initially proposed by Sedra and Smith in 1968. A modified version is presented as a second-generation current conveyor (CCII) in 1970. The CCII continues to receive much more attention, and finds numereous applications in many fields of electronics. This growing interest to the CC has led us to focus on this basic building block and its extensively used applications such as simulated inductors and active analog filters. Inductors are not desirable passive elements in most electronic circuits because the behavior of them is not very close to the ideal element behavior and they are physically larger and heavier than other components. In addition, it is very difficult to realize large-valued inductors in integrated circuit (IC) technology and maximum a few nH inductors with low Q can only be implemented with current technology. Instead of large-valued physical inductors, simulated inductors employing CCs or other active components are widely used in many analog circuits. Such inductors with capacitors and other components can be used to form tuned circuits to build filters. CC based circuits suffer from several restrictions such as parasitic impedance and frequency dependent non-ideal gain effects in addition to signal limitations stemmed from the restricted DC power supply voltages and limited allowable input currents. In fact, such restrictions also exist in other circuits realized with different active building blocks. The mentioned limitations usually result in some undesired conditions such as stability problems and distortions at the output terminals. In this thesis, the restrictions of the CC based current-mode and voltage-mode analog filters as well as simulated inductors that can be used as building blocks for analog filtering functions are investigated in detail. Throughout this thesis, some methods to overcome these restrictions have been proposed. A number of novel simulated inductors and analog filters with reduced parasitic impedance effects are presented as alternative circuits. Several experimental results using the commercially available active devices, AD844s are shown. Moreover, time and frequency domain as well as Monte Carlo analysis using SPICE and MATLAB programs are performed to verify the theory of the designed inductors and analog filter circuits.
Adaptive control of nonlinear systems using multiple identification models
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2007., 2007.) Cezayirli, Ahmet.; Cılız, Kemal.
Adaptive control of nonlinear systems is considered in this study. Available methods in this field are reviewed first. Focusing on the minimum-phase, input-output linearizable and linearly parameterized nonlinear systems, direct and indirect adaptive controllers are developed for the cases of matched and unmatched uncertainties. A new methodology is proposed, which makes use of multiple identification models in order to improve the transient performance under large parametric uncertainties. Adaptation and switching mechanisms are developed based on the use of multiple Lyapunov functions and cost functions. The resulting closed loop systems are shown to be stable with the switching mechanisms. Combination of direct and indirect adaptive control schemes is also presented for a class of nonlinear systems which do not give rise to over-parametrization. The theoretical results obtained in this study are verified by computer simulations.
Electronic tunability in analog filters
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2007., 2007.) Metin, Bilgin.; Çiçekoğlu, Oğuzhan.
In the beginning years of the analog design, discrete components were being used in the analog design procedures. After manufacturing, trimmer capacitors and potentiometers were used for electronic tunability to compensate for the electronic component nonidealities and tolerances. After 1970s, analog circuits have been implemented as integrated circuits, where element tolerances are unacceptably high. Non-idealities and parasitics due to IC implementation are serious problems for the circuits after manufacturing. Therefore, electronic tuning of some parameters after production has been a very important feature in IC design. In the thesis following major tunability methods are examined: Electronic tunability with MOSFET-C technique, with mixed translinear loops, with operational transconductance amplifiers and with adjusting current-gain. During electronic tuning of the desired parameters, some other parameters of the circuit such as linearity, stability, gain and high frequency performance, may be affected in an undesired way. Therefore, trade-offs between tunability and other parameters of the circuits are illustrated to find effective tunability ranges. Moreover, in the thesis new tunable filter circuits are suggested for each of these methods as examples and detailed non-ideality and parasitic component analyses of these circuits are given.
Sequential Bayesian modeling of non-stationary non-Gaussian processes
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2007., 2007.) Gençağa, Orhan Deniz.; Ertüzün, Ayşın.; Kuruoğlu, Ercan E.
This thesis brings a unifying approach for modeling non-stationary non-Gaussian signals which are widely encountered in many multidisciplinary research fields. In the literature, different approaches have been used to model non-stationary signals. However, they could not fulfill the increasing needs where non-Gaussian processes are involved until the development of Sequential Monte Carlo techniques (particle filters). In general particle filtering, the problem is expressed in terms of nonlinear and/or non-Gaussian state-space equations and we need information about the functional form of the state variations. In this thesis, we bring a general solution for cases where these variations are unknown and the process distributions cannot be expressed by a closed form probability density function. We propose a novel modeling scheme which is as unified as possible to cover these problems. First, a novel technique is proposed to model Time-Varying Autoregressive Alpha Stable processes where unknown, time-varying autoregressive coefficients and distribution parameters can be estimated. Successful performances have been supported by posterior Cramer Rao Lower Bound values. Next, we extend our methodology to model cross-correlated signals where vector autoregressive processes with non-Gaussian driving signals can also be modeled. Later, this extension is used as a building block to provide a more unifying solution where both mixing matrix and latent processes are modeled from their mixtures. This can be interpreted as a solution for non-stationary Dependent Component Analysis. Successful simulation results verify that our methodology is very flexible and provides a unifying solution for the modeling of non-stationary processes in all cases described above.
Cross-lingual voice conversion
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2007., 2007.) Türk, Oytun.; Arslan, Levent M.
Cross-lingual voice conversion refers to the automatic transformation of a source speaker’s voice to a target speaker’s voice in a language that the target speaker can not speak. It involves a set of statistical analysis, pattern recognition, machine learning, and signal processing techniques. This study focuses on the problems related to cross-lingual voice conversion by discussing open research questions, presenting new methods, and performing comparisons with the state-of-the-art techniques. In the training stage, a Phonetic Hidden Markov Model based automatic segmentation and alignment method is developed for cross-lingual applications which support textindependent and text-dependent modes. Vocal tract transformation function is estimated using weighted speech frame mapping in more detail. Adjusting the weights, similarity to target voice and output quality can be balanced depending on the requirements of the cross- lingual voice conversion application. A context-matching algorithm is developed to reduce the one-to-many mapping problems and enable nonparallel training. Another set of improvements are proposed for prosody transformation including stylistic modeling and transformation of pitch and the speaking rate. A high quality cross-lingual voice conversion database is designed for the evaluation of the proposed methods. The database consists of recordings from bilingual speakers of American English and Turkish. It is employed in objective and subjective evaluations, and in case studies for testing new ideas in cross- lingual voice conversion.
Finite element method with weighted extended b-splines for electromagnetics
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2007., 2007.) Apaydın, Gökhan.; Şeker, Selim.
This thesis presents fundamental overview of electromagnetics and finite element method (FEM) with weighted extended basis splines (web-splines), which is a new developed finite element method for electromagnetic problems. The developed method is discussed in detail. The advantages of FEM with web-spline method are illustrated by using several electromagnetic applications. The wave equation is solved by using web-spline method and compared with the previous studies in the literature. The results of the simulation are shown to have excellent agreements. Thus, this thesis proves that the FEM with web-spline method can be used in electromagnetic applications with good accuracy. The method does not need any mesh generation as in the standard FEM, and more accurate results are obtained by using less memory in computations.|Keywords: Electromagnetics; Error analysis; Finite element method; Waveguides
Density-based shape descriptors and similarity learning for 3D object retrieval
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2008., 2008.) Akgül, Ceyhun Burak.; Sankur, Bülent.; Schmitt, Francis.; Yemez, Yücel.
Next generation search engines will enable query formulations, other than text, relying on visual information encoded in terms of images and shapes. Content-based retrieval research aims at developing search engines that would allow users to perform a query by similarity of content. This thesis deals with two fundamentals problems in content-based 3D object retrieval: (1) How to describe a 3D shape to obtain a reliable representative for the subsequent task of similarity search? (2) How to supervise the search process to learn inter-shape similarities for more effective and semantic retrieval? Concerning the first problem, we develop a novel 3D shape description scheme based on probability density of multivariate local surface features. We constructively obtain local characterizations of 3D points on a 3D surface and then summarize the resulting local shape information into a global shape descriptor. This conversion mechanism circumvents the correspondence problem between two shapes and proves to be robust and effective. Experiments that we have conducted on several 3D object databases show that density-based descriptors are very fast to compute and very effective for 3D similarity search. Concerning the second problem, we propose a similarity learning scheme that incorporates a certain amount of supervision into the querying process. Our approach relies on combining multiple similarity scores by optimizing a convex regularized version of the empirical ranking risk criterion. This score fusion approach to similarity learning is applicable to a variety of search engine problems using arbitrary data modalities. In this work, we demonstrate its effectiveness in 3D object retrieval.
Advanced transceiver design for continuous phase modulation
(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2008., 2008.) Özgül, Barış.; Deliç, Hakan.; Koca, Mutlu.
This dissertation proposes advanced transceiver designs applying turbo and space-time (ST) concepts to continuous phase modulation (CPM), which is preferred in numerous power- and band-limited communication systems for its constant envelope and spectral e ciency. Despite its highly attractive spectral properties, maximumlikelihood detection of CPM over the frequency-selective multipath fading channels can bring impractical complexity issues because of the intensive search over a single super trellis which combines the e ects of the modulation and the multipath channel. Application of the reduced-state trellis search algorithms results in lower complexity but the computational load could still be prohibitively large to obtain high performance in long channel impulse responses. In the dissertation, instead of employing trellis-based combined detection methods, equalization and demodulation functions are separated and novel low-complexity receivers with soft-input soft-output (SISO) time-domain and frequency-domain linear equalizers are proposed for bit-interleaved coded CPM, which attain near-optimal performance by applying turbo processing. In the proposed receivers, the front-end soft-information-aided linear equalizer is followed by a central SISO CPM demodulator and a back-end SISO channel decoder where double turbo processing is employed by performing back-end demodulation/decoding iterations per each equalization iteration to improve the a priori information for the front-end equalizer. Performance for the frequency-domain equalization is further improved by proposing an orthogonal ST block coding scheme for CPM. The proposed technique maintains the constant envelope and the phase continuity of the CPM waveforms perfectly by using appropriate tail symbols and, therefore, has no impact on the spectral e ciency. Depending on the orthogonality of the ST combining, frequencydomain equalization is applied as in the case of single antenna transmissions without v any increase in the computational load. In the dissertation, the receiver complexity is reduced further by transferring all the equalization functions to the transmitter and employing pre-equalization. For precoding the CPM signals on multipath fading channels while maintaining the spectral e ciency, a novel ST pre-equalizer is proposed, limiting the envelope variations and attaining a peak-to-average power ratio that is close to one by using a transmit selection diversity scheme.

Browse

Browsing Ph.D. Theses by Issue Date

Results Per Page

Sort Options