No description available.

REROUTE SEQUENCE PLANNING IN MULTIPROTOCOL LABEL SWITCHING NETWORKS &Ouml / zbilgin, izzet G&ouml / khan M.S., Department of Electrical and Electronics Engineering Supervisor: Asst. Prof. Dr. C&uuml / neyt F. Bazlama&ccedil / ci December 2004, 57 pages The growth of the Internet has caused the development of new protocols that enable IP networks to be engineered efficiently. One such protocol, Multiprotocol Label Switching (MPLS) enables IP datagrams in backbone networks to be forwarded based on the label switching forwarding paradigm. In MPLS networks, rerouting of Label Switching Paths (LSPs) can be needed in order to attain a better resource utilization in the network. In this case, a sequence of LSPs has to be found for their one by one reconfiguration without service interruption, involving the constraint that the link capacities should not be violated at any time during the rerouting process. This reroute sequence planning problem for LSPs is NP-complete. In previous works, the conditions of existence of any feasible reroute sequence are examined and algorithms are described for solving the problem, but it was shown that the problem is computationally hard in real-world situations because of the large amount of routers and LSPs in the network. In this work, we deal with the problem of reroute sequence planning problem of LSPs and present alternative solutions for the case when there is no feasible solution. We introduce a tool for the post-processing phase when a capacity violation is occurred during the sequence planning. We present an algorithm trying to reconfigure LSPs while allowing some interruption or degradation of traffic during the rerouting process.

The study of matrices have been of interest to mathematicians for some time. Recently the use of matrices has assumed greater importance also in the fields of management, social science, and natural science because they are very useful in the organization and presentation of data and in the solution of linear equations. The theory of matrices is yet another type of mathematical model which we can use to solve many problems that arise in these fields. The aim of this paper is to show how we can use matrices and their mathematical model to solve some problems in the process of routing. First we will introduce the term of routing and the new approach in the process of selecting paths. We will show some simple examples. We will also pint out how we can learn about matrices in the classroom. At the end we will discuss about advantages and potential disadvantages that may occur in the described technique.

The World Wide Web contains rich collections of digital materials that can be used in education and learning settings. The collaborative authoring prototype of Walden's Paths targets two groups of users: educators and learners. From the perspective of educators, the authoring tool allows educators to collaboratively build a Walden's Path by filtering and organizing web pages into an ordered linear structure for the common information needs, which can be extended, tailored and modified into a derivative path from its parent version to meet dynamic and evolving educational requirements. From the students' perspective, Walden's Paths provide a shared knowledge space that facilitates collaborative learning. Specifically, collaborative learners can annotate locally and globally on pages and share among group members, where each annotation fosters the initiation of a thread of discussion. Therefore, knowledge transfer can be achieved in the process of social interaction associated with shared annotations.

Thesis (M.S.)--Rochester Institute of Technology, 2008. / Typescript. Includes bibliographical references (leaves 64-66).

Thesis (M.U.D.)--University of Hong Kong, 1996. / Includes bibliographical references (leaves 75-79) Also available in print.

The past three decades have seen extraordinary progress in the manipulation of neutral atoms with laser light, to the point where it is now routine to trap and cool both individual atoms and entire atomic clouds to temperatures of only a few tens of nanoKelvin in a controlled and repeatable fashion. In this thesis we study several applications of Rydberg atoms - atoms with an electron in a highly excited state - within such ultracold atomic systems. Due to their highly-excited electron, Rydberg atoms have a number of exaggerated properties: in addition to being physically large, they have long radiative lifetimes, and interact strongly both with one another and with applied external fields. Rydberg atoms consequently find many interesting applications within ultracold atomic physics. We begin this thesis by analysing the way in which a rubidium atom prepared in an excited Rydberg state decays to the ground state. Using quantum defect theory to model the wavefunction of the excited electron, we compute branching ratios for the various decay channels that lead out of the Rydberg states of rubidium. By using these results to carry out detailed simulations of the radiative cascade process, we show that the dynamics of spontaneous emission from Rydberg states cannot be adequately described by a truncated atomic level structure. We then investigate the stability of ultra-large diatomic molecules formed by pairs of Rydberg atoms. Using quantum defect theory to model the electronic wavefunctions, we apply molecular integral techniques to calculate the equilibrium distance and binding energy of these molecular Rydberg states. Our results indicate that these Ryberg macro-dimers are predicted to show a potential minimum, with equilibrium distances of up to several hundred nanometres. In the second half of this thesis, we present a new method of symbolically evaluating functions of matrices. This method, which we term the method of path-sums, has applications to the simulation of strongly-correlated many-body Rydberg systems, and is based on the combination of a mapping between matrix multiplications and walks on weighted directed graphs with a universal result on the structure of such walks. After presenting and proving this universal graph theoretic result, we develop the path-sum approach to matrix functions. We discuss the application of path-sums to the simulation of strongly-correlated many-body quantum systems, and indicate future directions for the method.

A busca de soluções para a mobilidade urbana que minimizem a agressão do setor de tráfego e transportes ao meio ambiente está cada vez maior. Os veículos elétricos se posicionam como uma alternativa interessante, pois reduzem a emissão de gases poluentes na atmosfera, a poluição sonora e o consumo de petróleo. No entanto, sua limitada autonomia e a escassez de postos de recarga intimidam sua adoção. Por conta disso, políticas governamentais de incentivo têm sido desenvolvidas para a oferta de benefícios a quem optar por um veículo elétrico. Estima-se que dentro de poucas décadas toda a frota urbana será substituída por veículos dessa natureza. Por isso, é importante entender as mudanças no tempo de viagem e no consumo de energia oriundos da inclusão de veículos elétricos em cenários de tráfego. Trabalhos anteriores estudaram as diferenças entre os mecanismos internos de veículos convencionais e elétricos na determinação destas mudanças. Porém, dadas as características destes últimos, motoristas de veículos elétricos se preocupam com a economia de energia e podem optar por rotas diferentes. Logo, uma análise completa destes impactos deve considerar uma nova distribuição de tráfego. Este trabalho propõe um método biobjetivo de alocação de tráfego que considera o tempo de viagem e o consumo de energia para determinar a distribuição de veículos elétricos em cenários de tráfego urbano. Duas estratégias de distribuição de fluxo são propostas como mecanismos de escolha de rotas. Como parte da alocação de tráfego, é proposto um algoritmo biobjetivo de caminhos mínimos para veículos elétricos. A abordagem apresentada foi aplicada a três cenários distintos, onde percebeu-se uma diminuição de até 80% no consumo total de energia. Em cenários com congestionamento, observou-se um aumento de 10% no tempo de viagem. Já em cenários sem congestionamento o tempo de viagem diminuiu cerca de 2%. A recuperação de energia representa quase 6% da economia total dos veículos elétricos. Além disso, experimentos mostraram que investimentos na eficiência dos veículos elétricos podem resultar em uma economia de até 15% de energia. / The search for urban mobility solutions that minimize the aggression to the environment is increasing. Electric vehicles are an attractive alternative because they reduce greenhouse gas emissions, noise pollution, and oil consumption. However, their limited autonomy and the lack of charging stations restrict their popularization. Therefore, government incentive policies have been developed in order to offer benefits to those who choose an electric vehicle. It is estimated that the entire urban fleet will be replaced by these vehicles in a few decades. Therefore, it is important to understand the changes in travel time and energy consumption from the inclusion of electric vehicles in traffic scenarios. Previous works determined these changes by studying the differences between the internal engine of conventional and electric vehicles. However, given the characteristics of the latter, drivers of electric vehicles care about saving energy and may want to choose different routes. Thus, a complete analysis of these impacts should consider a redistribution of traffic. This work proposes a bi-objective traffic assignment method that considers the travel time and the energy consumption to determine the distribution of electric vehicles in urban traffic scenarios. We introduce two strategies for flow distribution as models of route choice. As a procedure of the traffic assignment method, we propose a bi-objective shortest path algorithm for electric vehicles. Our approach was applied to three different scenarios, which resulted in a decrease of up to 80% in total energy consumption. In congested scenarios, we observe an increase of about 10% in average travel time. In uncongested scenarios, travel time decreases about 2%. Energy recovery is almost 6% of the total savings of electric vehicles. Moreover, experiments have shown that investments in the efficiency of electric vehicles can result in up to 15% of energy savings.

Algoritmos de similaridade que comparam dados expressos em XML são importantes em diversas aplicações que manipulam informações armazenadas nesse padrão. Sistemas de integração de dados XML e de consulta a instâncias XML são exemplos dessas aplicações. A utilização de funções de similaridade para efetuar as comparações nessas aplicações melhora seus resultados finais. A melhora ocorre porque as funções de similaridade possibilitam encontrar estruturas não idênticas às apresentadas nos parâmetros das consultas mas que armazenam informações relevantes. Uma característica importante que pode ser utilizada para definir se dois elementos XML representam o mesmo objeto real é os caminhos que chegam a estes elementos nas suas respectivas árvores. No entanto, os nodos que representam um determinado objeto real em duas instâncias XML diferentes podem se acessados por caminhos distintos, devido a opções de modelagem dos documentos. Portanto um algoritmo para calcular a similaridade entre caminhos XML é importante para as aplicações descritas acima. Neste contexto, esta dissertação objetiva desenvolver um algoritmo de similaridade entre caminhos XML. O resultado principal do trabalho é um algoritmo de similaridade entre caminhos XML, nomeado PathSim, que efetua o cálculo de similaridade entre dois caminhos baseado no número mínimo de operações de edição (inserção, remoção e substituição de nomes de elementos) necessárias para transformar um caminho no outro. Além deste algoritmo, foram desenvolvidas três funções de pré-processamento para simplificar os caminhos XML e melhoram os resultados do algoritmo. Adicionalmente, duas variações do algoritmo PathSim são apresentadas, uma incrementada com comparações entre combinações de nomes de elementos, nomeada PathSimC, e a outra auxiliada por técnicas de alinhamento, nomeada PathSimA. Experimentos utilizando documentos XML criados por terceiros, validam empiricamente os algoritmos PathSim e PathSimC.Nos experimentos, os algoritmos foram comparados a uma abordagem para mensurar a similaridade entre caminhos encontrada na literatura. Os algoritmos apresentam melhores resultados que o baseline. Os ganhos variam de acordo com o ambiente onde os caminhos foram extraídos e com as funções de pré-processamento que foram aplicadas aos caminhos. / Similarity algorithms for comparing XML data are important in various applications that manipulate information stored according to this standard. XML data integration systems and XML instance querying systems are examples of such applications. The use of similarity functions to evaluate comparisons in these applications improves their final results. The improvement occurs because similarity functions allow finding structures that are not identical to the query parameter but store relevant information. One important feature that may be used to define if two XML elements represent the same real world object is the paths that lead to those objects in their corresponding trees. However, the nodes that represent a specific real world object in two different XML instances may be accessed by distinct paths, due to XML design decisions. Thus a method for assessing the similarity of XML paths is important in the applications described above. In this context, the goal of this dissertation is to develop a XML path similarity algorithm. The main contribution of this work is a XML path similarity algorithm, named Path- Sim, that calculates the similarity between two paths by computing the minimum number of edit operations (element name insertions, deletions and substitutions) required to transform one path into another. Besides the algorithm, three preprocessing functions were developed to simplify XML paths and improve the results of the algorithm. Additionally, two variations of PathSim algorithm are presented, one enhanced with comparisons among combinations of element names, named PathSimC, and the other one assisted by alignment techniques, named PathSimA. Experiments using XML documents created by third parties validate the algorithms PathSim and PathSimC empirically. On the experiments, the algorithms are compared to a path similarity algorithm found in the literature. The proposed algorithms presents better results than the baseline. The gains vary according to the environment from which the paths were extracted and to the preprocessing functions applied.

XML documents and related technologies represent one of the most widespread ways how data on the Web are maintained and interchanged. Unfortunately, many of the real-world documents contain various types of consistency issues that prevent their successful automated processing. In this thesis we focus on the problem of the structural invalidity and its correction. In particular, having one potentially invalid XML document modeled as a tree, and a schema in DTD or XML Schema languages modeled as a regular tree grammar, our goal is to find all the minimal corrections of this tree. The model we proposed builds on top of the recursively nested structures of correction multigraphs, where the shortest paths are being found. For this purpose we formally introduce three correction strategies with different pruning optimizations applied. According to the experiments we performed, the refinement correction strategy not only significantly outperforms all the other existing approaches, but also guarantees important characteristics the others cannot. Powered by TCPDF (www.tcpdf.org)