Institute of Graduate Studies in Science

Object relocation in a mobile computing environment

MaROS is an application development platform, which offers a solution to strengthen the weak aspects of mobile systems such as low network bandwidth, limited disk space, limited processing power, and batteries with short life span. In this platform, registered mobile computers may transfer their applications to MaROS server. This process is called "relocation". In this thesis, the design and implementation of the MaROS migration module is presented. The proposed model supports a lightweight migration system initiated by the application. A brief survey on migration systems, design, implementation, and evaluation of entire MaROS system and its migration package are presented in this thesis. In addition, suggestions to enhance the system are presented.

Aggregate production planning by linear programming with an application

Tedarik zinciri yönetimi için grup karar destek sistemleri

ABSTRACT As organizations continously seek to provide their products and services to customers faster and cheaper, managers have come to realize that they can not accomplish it alone; they hence must work on a cooperative basis in Supply Chain Management (SCM). Nowadays, many important decisions in organizations are made not only by a single individual, but also by groups of individuals. Group Decision Support Systems (GDSS) enable organizations to focus on key business activities especially in Supply Chain Management Process. In this process, from the raw material to the end customer there are two flows, which are the material flow and the information flow. These flows must be managed successfully via GDSS in SCM. In every stage of SCM, there is a selection process. Therefore, the firms may handle this selection by effectively using GDSS. In GDSS, there is a multi criteria decision aid. This aid aims to give the decision maker some tools and techniques in order to enable him to advance in solving a decision problem where several points of view must be taken into account. Therefore, Multi Criteria Decision Making (MCDM) methods have received increased attention and become a practical approach for certain types of problems in the real life. This research describes the development of a model for the selection of laptop computers. The purpose of this study is to apply Multiple Attribute Decision Making (MADM) methods to the selection of laptop computers for a firm (which is the procurement stage of SCM) by using GDSS. Our model consists of 279 different types of laptop computers and 86 criteria. The methods used in this study are Analytic Hierarchy Process (AHP), REMBRANDT, ELECTRE, and PROMETHEE. The following steps were carried out while applying the methods to the problem. Firstly, the hierarchy tree was created by decomposing into criteria and sub-criteria.. Secondly, the weigths for the criteria and alternatives were determined. The relevant weights were calculated using pairwise comparisons by GDSS. Finally MCDM methods were applied to this model. Programming language Java was used to implement the considered methods. The results look like similar. Based on this model, the research provides a means whereby companies or organizations can assess the selection of activities, which are critical to their business by using GDSS. Keywords: Group Decision Support Systems (GDSS), Supply Chain Management (SCM) Multi Criteria Decision Making (MCDM), Multiple Attribute Decision Making (MADM), AHP, REMBRANDT, ELECTRE and PROMETHEE Methods

A Mathematical model for planning of class timetables an application in Turkish Military Academy

A timetabling problem of an educational institution involves scheduling a large number of classes, teachers and courses to a number of periods while satisfying a set of constraints of various types. The aim of this study is to find good and satisfactory solutions for the timetabling problem of Turkish Military Academy in shorter time when compared to the manual timetabling process. An integer programming model which has the appropriate facilities for providing valuable help to the scheduler to implement a good timetable of Turkish Military Academy and a solution strategy for this model is presented. The model satisfies the teachers' preferences to a sufficient degree by using suitable objective function coefficients. These coefficients penalize the assignment of teachers to the periods which they do not prefer. They also enable us to assign the part-time teachers at their available periods prior to the full-time ones. In addition, the solution strategy proposed is based on grouping sections and teachers, and on allocating teacher groups to section groups iteratively in such a way that the output data of an iteration will be added to the input data of the next one. This grouping strategy made it possible to decompose the problem into the problems of smaller sizes. The solution found by the implementation of the model in this study satisfies all problem requirements and minimizes the objective function. It is near- optimal within the iteration limits and is found in several hours. The timetable constructed has only one non-preferred period for only one teacher. Besides, the time consumed for this model is very short when compared with the manual timetabling. Thus, it is said to be "a good timetable" both for the teachers and the scheduler. IX

A data mining approach to predict failures in banking sector

Proje risklerinin sistem dinamik modellemesi

ABSTRACT Risk, defined as exposure to the possibility of economic or financial loss or gains, physical damage, or injury, or delay as a consequence of the uncertainty associated with pursuing a cause of action, is by nature subjective. The fast changing environmental conditions, information technology, and the complexity of the projects increase risk threat and make project management more susceptible to risks. This sensitiveness is also at its highest level in telecommunication projects, because there are various technologies and scientific techniques used in telecommunication projects increasing the uncertainty and complexity. Project risk dynamics are difficult to understand and control. Therefore managing project risks requires an approach supported by special tools and techniques. System dynamic (SD) modeling is a tool that covers a wide range of project management needs, by addressing the system issues that influence and often dominate the project outcome. With SD modeling approach we will be able to see all things as a whole. SD modeling also will enable us to build formal computer simulation of complex project risk dynamics, so we can identify and analyze probable risks within the project and respond to them. In the thesis project, we attempt to classify all project risks within the framework of PSO (People-System-Organization) concept. Five risk sectors are defined. 1. Social Risk 2. Economic Risk 3. Technical Risk 4. Political Risk 5. Ecological Risk Political and Ecological risk sectors are excluded and the first three sectors are defined as endogenous in the model. In Stock and Flow Map these three sectors are determined as main stocks. Modeling these three risk sectors stock and flow diagrams 46 risk variables are used. There are three main stocks and seven sub-stocks in the model. Moving from causal map to computer simulation we used the approach called NUMBER (Normalized Unit Modeling by Elementary Relationships), which enable us to limit the variable's values between 0 and 1. We needed to use this approach because we did not have the real data. After forming the equations we simulated the model with time bound which is assumed to be 60 months. Simulation results show that all main risk sectors increase in the initial times of the project. Then if all thins go well the risk levels begin to decrease but never zero. Economic Risk shows the highest level about 0.6 at the time 18 in the base model The highest levels of the other two main risk factors are about 0.3 at a determined time. Near the base model we defined an alternative model to which a new variable called Hiring New Staff is added. After simulating the alternative model it was observed that to introduce new personnel to an ongoing project increases risks levels and project cost, decreases productivity, quality and leads to a poor team relationship. The model can be simulated with different policies and designs. It enables us to see various situations of risk factors and other variables affecting risks. In this way, irrelevant risks can be eliminated, preventing unnecessary mitigating efforts. At the end of this study we are able to have a complete picture of telecommunication project risk dynamics with the help of an SD model computer simulation. This may help us in seeing risk behaviors during project time and formulating responses to risks for mitigating their impacts on telecommunication projects. XIV

Solutions of the navier-stokes equations and technological applications using matrix distribution with N-methods

IV ABSTRACT SOLUTIONS OF THE NAVTER-STOKES EQUATIONS AND TECHNOLOGICAL APPLICATIONS USING MATRIX-DISTRIBUTION WITH N-METHODS The interaction of an externally imposed magnetic and electric field on the laminar flow of a conducting fluid in a channel is studied using computational techniques. The Navier-Stokes (N-S) equations and the equations describing the electromagnetic field (MHD) are solved simultaneously in a computational fluid dynamics code that utilizes a matrix distribution scheme. The flow considered is two-dimensional on x-y plane. The MHD flow includes an imposed electric and magnetic field acting also on this plane along the central region of the channel and activated by a switch at a specified time (after the vertical flow reaches steady state). A magnet is placed on both sides of the channel and an external magnetic field configuration is obtained in the direction connecting the left and right walls. In addition, an external electric field is applied in perpendicular direction such that ExB becomes opposite to gravitational acceleration direction. This is done to slow down (or even control) the vertical velocity in the channel. The calculations are done with a new solver using a matrix distribution scheme that works on structured and unstructured triangular meshes. The N-S and MHD equations are first written in nondimensional forms and the numerical scheme was obtained. For time iterations, a 3-stage Runga-kutta method is used. The numerical results show that this computer code is capable of resolving different types of problems in different geometries.

Layered microstrip helical antenna design

IV ABSTRACT LAYERED MICROSTRIP HELICAL ANTENNA DESIGN A novel broadband microstrip antenna has been developed to advance basic characteristics of microstrip antennas such as narrow bandwidth and low directivity. Patches which have the geometrical configuration of annular sector type are arranged on layered media. The diameters of the patches become smaller into the upper layers of the designed microstrip antenna. The active patches are connected to each other by a short transition wire; therefore the geometrical configuration of the novel antenna seems to be a tapered helical antenna. Analytical models and numerical techniques are applied to find the fundamental antenna parameters for an annular sector microstrip patch antenna, and the results are compared to each other. The analytical techniques, with some appropriate assumptions, based on superposition and array theorems in electromagnetics are applied for layered microstrip helical antenna design, and the results of the fundamental antenna parameters are compared with the results of those of uniform and quasi-tapered helical antennas, approximately. The study shows that the layered microstrip helical antenna increases the pattern bandwidth and the directivity of the antenna, and also reduces half- power beamwidth, as it is expected. The layered microstrip helical antenna can be analyzed by using Dyadic Green's functions to find out more accurate results. This method is complex to apply, therefore the mentioned approximate analytical models have been developed and applied.

Hand gesture recognition using artifical neural networks

ABSTRACT HAND GESTURE RECOGNITION USING ARTIFICIAL NEURAL NETWORKS In order for humans to interact with computers, a fast and easy way is to use hand gestures. Although using hand gestures in computer interaction was very cumbersome and needed special glove and computer hardware in the past, nowadays in any personal computer, simple cameras are available and there is enough processor power to do the expensive computations done in the past. With this feature, not only controlling computer with hand is possible but also some simple interpreters for sign language recognition can be made. In this thesis a hand gesture recognition system, using an inexpensive camera and a personal computer is proposed. The system uses visual image as input and computes the geometric features, such as invariant moments and signature, of the extracted hand image for classification in a Multi-layer Perceptron Artificial Neural Network. The gestures used in the system are American Sign Language Manual Alphabet Gestures and Turkish Sign Language Manual Alphabet Gestures.

Isomorphism theorems of linear groups

Genco FASThe goal in this thesis is the isomorphism theory of linear groups over fields as shownby the theoremTheorem: There is an isomorphism between any subset of the (projective) collineartransformations, (projective) general linear groups or (projective) special linear groups fordimension ≥ 3 if and only if the dimensions of the finite dimensional vector spaces of thesesubspaces are equal and the underlying fields of these vector spaces are isomorphic.The theory that follows is typical of much of the research between the years 50?s and60?s on the isomorphisms of the classical groups over rings. The thesis will start from thebasic facts of calculus of residues and transvections. Then, in particular, the fundamentaltheorem of projective geometry will be proved and whatever is needed from projectivegeometry will be developed. Via reorganizing the literature on the isomorphisms of theclassical groups, it will be possible to extend the known theory from groups of lineartransformations to groups of collinear transformations, and also to improve the isomorphismtheory from dimension ≥ 5 to dimension ≥ 3.