La localisation de source d'EEG devient un outil important pour traiter les patients atteints d'épilepsie en localisant les zones épileptogènes avant d'effectuer une chirurgie de résection. Compte tenu d'un modèle de tête direct, la localisation de la source EEG est réalisée en résolvant le problème inverse. Le modèle de tête direct est un modèle biophysique de tête plus ou moins complexe qui décrit la distribution électrique. En considérant la propagation électrique expliquant la distribution de potentiels, outre la numérisation, le modèle nécessite le réglage deux paramètres lesquels sont la géométrie du modèle de tête et la valeur des conductivités de chaque compartiment du modèle de tête. En raison des progrès computationnel et des techniques d'imagerie (comme l'IRM et la CT), il est possible de générer des modèles de tête humaine qui représentent avec une grande précision la géométrie de la tête réelle. Cependant, il existe une incertitude sur les valeurs de conductivité de chaque compartiment et la méthode avec laquelle ils devraient être estimés. Dans la littérature, les valeurs communes pour les conductivités proviennent principalement des expériences in-vitro. Dans ce travail, nous effectuons une estimation de la conductivité in-vivo à partir de données EEG/SEEG/Stimulation électrique de trois patients épileptiques. Ces données sont constituées des images IRM et des CT SCAN pour la construction d'un modèle de tête FEM à cinq compartiments pour chaque patient, ainsi que les enregistrements SEEG et EEG qui ont été acquis en même temps que la stimulation électrique intracérébrale (IES). L'originalité de ce travail réside dans l'évaluation de la performance de l'estimation des conductivités in-vivo par des mesures EEG et / ou SEEG en fonction de différents paramètres spatiaux et de la localisation des IES. Le travail se compose de trois parties principales: la première partie vise à déterminer la méthode d’optimisation sous contraintes la plus robuste parmi les algorithmes courants pour optimiser les paramètres du modèle direct de tête. L'objectif de la deuxième partie est d'analyser la sensibilité des valeurs de conductivité à différentes conditions sur la position de stimulation, le conditionnement du problème avec les positions de mesure et leur nombre et le nombre de compartiments. Alors que dans la partie finale, les conductivités d'un modèle de tête FEM isotrope et homogène à cinq compartiments ont été estimées avec des paramètres précédemment déterminés pour les trois patients. Enfin, l'effet de la fréquence de stimulation sur les conductivités estimées est analysé / EEG source localization is becoming an important tool for treating epileptic patients by localizing the epileptogenic zones before performing a resection surgery. Given a forward head model, EEG source localization is performed by solving the inverse problem. The forward head model is a biophysical model which describes the electrical distribution in the human head. When considering the propagation as the only way for the current distribution to move in the head, the focus is directed primarily on two parameters for having an accurate forward head model. These parameters are: the geometry of the head model and the conductivity value of each compartment of the head model. Due to the recent advances in computers and imaging techniques (like MRI and CT), it is possible to generate human head models that represent with a high accuracy the geometry of the real head. However, there is still an argument about the conductivity values and the method by which it should be estimated. In literature, the common values for conductivities come mostly from in-vitro experiments. In this work we are performing in-vivo conductivity estimation by considering the data of three epileptic patients. This data consists of MR images and CT scans for building a five-compartment FEM head model for each patient along with SEEG and EEG recordings that were acquired in simultaneous with intracerebral electrical stimulation (IES). The originality of this work lies in evaluating the performance of in-vivo conductivity estimation by EEG and/or SEEG measurements in function of different spatial parameters and locations of the IES. The following work consists of three major parts: the first part aims to determine the most robust optimization algorithm among common algorithms for optimizing the forward head model. The objective of the second part is to analyze the sensitivity of the conductivity values given different conditions on stimulation position, measurement positions and number of compartments. While in the final part, the conductivities of an isotropic and homogeneous five-compartment FEM head model were estimated with previously selected parameters for three drug-resistant epileptic patients. Finally the effect of changing the stimulation frequency on the estimated conductivities was determined

Em telecomunicações é utilizada uma grande variedade de modelos de propagação para prever a intensidade do sinal de recepção num enlace de rádio. Os resultados dos modelos empíricos e dos modelos baseados em simplificações do problema eletromagnético apresentam, em geral, grande discrepância entre si, na análise do fenômeno de propagação. Por exemplo, o modelo de dois raios, que prevê atenuação de enlace com variação de 40 dB/década, na região de campo distante da antena de transmissão, quando colocada a uma certa altura da superfície de um solo condutor perfeito. Já o modelo empírico de Okumura-Hata exibe, para as mesmas condições geométricas, mas para solo de condutividade finita, uma variação da atenuação de enlace de 30 dB/década. Neste trabalho propõe-se uma calibração do modelo de dois raios para aumentar sua aderência aos resultados obtidos por Okumura-Hata num cenário real na faixa de frequências de 450MHz a 1900MHz para comprimentos de enlace entre 1 km e 20 km. Os resultados obtidos com o modelo proposto são comparados com os obtidos através de outros modelos apresentando boa aderência verificada através de uma comparação numérica particularmente com o modelo Okumura-Hata. / In telecommunications, there is a large variety of propagation models in order to predict the intensity of the receiving signal in a communication link. The empirical models and the models based on simplifications of the electromagnetic problem with theoretically obtained results show, in general, a significant disagreement, for the same propagation phenomenon. For example two-ray model predicts a path-loss of 40 dB/decade in the far-field region of a transmission antenna deployed at a certain height of the surface of a perfect conducting soil. On the other hand, the Okumura-Hata empirical model shows, for the same geometrical conditions, bur for a real soil, a path-loss of 30 dB/decade. This work proposes a calibration of the Two-Ray model for better adherence to the results obtained by Okumura-Hata in a real scenario for the frequency range 450MHz to 1900MHz and for a radio link length of 1 km to 20 km. The results obtained with the proposed model are compared with other model results showing good adherence through numeric comparison, in particular with the Okumura - Hata model.

Submitted by Ricardo Cedraz Duque Moliterno (ricardo.moliterno@uefs.br) on 2017-07-11T22:36:03Z No. of bitstreams: 1 Rafael Tosta Santos - Dissertacao Final.pdf: 20325954 bytes, checksum: bc011c09cd83ca73255a8b092923400a (MD5) / Made available in DSpace on 2017-07-11T22:36:03Z (GMT). No. of bitstreams: 1 Rafael Tosta Santos - Dissertacao Final.pdf: 20325954 bytes, checksum: bc011c09cd83ca73255a8b092923400a (MD5) Previous issue date: 2016-05-31 / This master thesis presents a model of deterministic propagation for waveguides with transmitting antenna above the top of the buildings and the equation for the diffraction at the top of the buildings. From the model developed by Dr. Edgar Silva Jr., adaptations are made to consider buildings of different heights along a waveguide. It is developed especially for this work calculations for diffraction at the top for a more accurate model. All the equipment and software used in the process of collection and preparation of the measured data, the sites used for model validation in the city of Feira de Santana - BA and S?o Paulo - SP, in Brazil, the construction of the 3D model and extraction sites database using Google Earth are described. Comparisons with measured values are show that the developed model can be a useful tool in signal envelope behavior analysis for waveguides with antennas above the top of the buildings. / Este trabalho de mestrado apresenta um modelo de propaga??o determin?stico para guias de ondas com antena transmissora acima do topo dos pr?dios e o equacionamento para as difra??es no topo dos pr?dios. A partir do modelo desenvolvido pelo Dr. Edgar Silva J?nior, s?o realizadas adapta??es para considerar constru??es de diferentes alturas ao longo de um guia de ondas. Sendo desenvolvido especialmente para este trabalho c?lculos para difra??o no topo para maior precis?o do modelo. S?o descritos todos os equipamentos e softwares utilizados no processo de coleta e prepara??o dos dados medidos, os locais utilizados para a valida??o do modelo na cidade de Feira de Santana - BA e S?o Paulo - SP no Brasil, a constru??o do modelo 3D e extra??o da base de dados dos locais utilizando o Google Earth. As compara??es com valores medidos mostram que o modelo desenvolvido pode ser uma ferramenta ?til na an?lise do comportamento do sinal para guias de ondas com antenas acima do topo dos pr?dios.

Made available in DSpace on 2016-04-04T18:31:23Z (GMT). No. of bitstreams: 1 KERLLA DE SOUZA LUZ OLIVEIRA.pdf: 1568205 bytes, checksum: bce8e961721a4346ba548c8f58ba336a (MD5) Previous issue date: 2008-05-29 / This work evaluates the coverage of a WMAN (Wireless Metropolitan Area Network), with characteristics of a system WiMAX (Worldwide Interoperability will be Microwave Access) operating in 5.8GHz, in accordance with the used modulation. In order to reach this goal, predictions of field covering were used and measurements were made. The modulation of a adaptative antenna depends on the value of the received power and the attenuation of this power varies according to the region where the signal propagates. In this work the distribution of the modulation in the coverage area was evaluated using the Shadowing model to describe the attenuation of the signal in accordance with the considered environment. As a result, it could be observed the association of areas with two types of possible modulation due to the overlap of modulation bands generated by the standard deviation of the signal medium value. / Este trabalho avalia a cobertura de uma WMAN (Wireless Metropolitan Area Network), com caracter?sticas de um sistema WiMAX (Worldwide Interoperability for Microwave Access) operando em 5.8GHz, de acordo com a modula??o utilizada. Para tanto, empregaram-se tanto predi??es de cobertura quanto medidas de campo. A modula??o de uma antena depende do valor da pot?ncia recebida e a atenua??o desta pot?ncia varia de acordo com a regi?o por onde o sinal se propaga. Neste trabalho foi avaliada a mudan?a da modula??o dentro da ?rea de cobertura, utilizando-se o modelo de Shadowing para descrever a atenua??o do sinal de acordo com o ambiente considerado. Como resultado p?de-se observar a associa??o de ?reas com dois tipos de modula??es poss?veis, devido ?s faixas de sobreposi??o de modula??o geradas pelo desvio padr?o do valor m?dio do sinal.

Em telecomunicações é utilizada uma grande variedade de modelos de propagação para prever a intensidade do sinal de recepção num enlace de rádio. Os resultados dos modelos empíricos e dos modelos baseados em simplificações do problema eletromagnético apresentam, em geral, grande discrepância entre si, na análise do fenômeno de propagação. Por exemplo, o modelo de dois raios, que prevê atenuação de enlace com variação de 40 dB/década, na região de campo distante da antena de transmissão, quando colocada a uma certa altura da superfície de um solo condutor perfeito. Já o modelo empírico de Okumura-Hata exibe, para as mesmas condições geométricas, mas para solo de condutividade finita, uma variação da atenuação de enlace de 30 dB/década. Neste trabalho propõe-se uma calibração do modelo de dois raios para aumentar sua aderência aos resultados obtidos por Okumura-Hata num cenário real na faixa de frequências de 450MHz a 1900MHz para comprimentos de enlace entre 1 km e 20 km. Os resultados obtidos com o modelo proposto são comparados com os obtidos através de outros modelos apresentando boa aderência verificada através de uma comparação numérica particularmente com o modelo Okumura-Hata. / In telecommunications, there is a large variety of propagation models in order to predict the intensity of the receiving signal in a communication link. The empirical models and the models based on simplifications of the electromagnetic problem with theoretically obtained results show, in general, a significant disagreement, for the same propagation phenomenon. For example two-ray model predicts a path-loss of 40 dB/decade in the far-field region of a transmission antenna deployed at a certain height of the surface of a perfect conducting soil. On the other hand, the Okumura-Hata empirical model shows, for the same geometrical conditions, bur for a real soil, a path-loss of 30 dB/decade. This work proposes a calibration of the Two-Ray model for better adherence to the results obtained by Okumura-Hata in a real scenario for the frequency range 450MHz to 1900MHz and for a radio link length of 1 km to 20 km. The results obtained with the proposed model are compared with other model results showing good adherence through numeric comparison, in particular with the Okumura - Hata model.

在資訊來源日趨複雜化及多樣化之下，過多不必要的資訊反而造成決策上的困擾，因此資訊的篩選(Information Filtering)便成為設計資訊系統時所要考量的重要因素之一。資訊的有效篩選不僅使得決策的不確定性降低，同時讓決策人員能夠專注於對決策有重要影響的因素上，提高了決策的效率與品質。本研究即是以逆傳遞(Back Propagation)類神經網路模式(Artificial Neural Network Model)為基礎，設計一個能夠篩選出重要屬性的通用演算法；此演算法能夠幫助使用者去除一些對決策較無影響的屬性，讓使用者能夠減少資訊收集成本，並針對重要屬性做決策上的考量。同時在本研究中，我們還將此演算法應用在生產現場的屬性篩選上，幫助排程人員找出對於排程法則選取有重要影響的屬性；並藉此驗證篩選演算法的正確性及完整性。 / To screen the mformation effectively can improve the efficiency and quality of decision making dramatically. Since it does not only decrease the uncertainty of decision maldng, but also let decision makers can emphasize on the important factors which can significantly affect the result of decision. In this thesis we present an algorithm to find the important factors out based on the technique of back propagation neural network model. This algorithm can help users to remove some attributes which do not or seldom affect the result of decision, and let them can reduce the cost of data collection and emphasize on consideration of the remaining important attributes. And in this thesis, we also apply the algorithm to filter out the important production attributes of shop floor scheduling system which can significantly affect the selection of shop floor scheduling rules, and use the result of this experiment to verify the correctness and completeness of the algorithm.

Indoor/Outdoor Location of Cellular Handsets Based on Received Signal Strength Jian Zhu 123 pages Directed by Dr. Gregory D. Durgin Accurate geo-location is an important emerging technology for public safety, commercial use, and military application. Especially, in the United States, the wireless Enhanced 911 (E911) rules by the Federal Communication Commission (FCC) seek to provide 911 dispatchers with additional information on wireless 911 calls. This dissertation presents a novel technique for indoor/outdoor location of cellular handsets based on received signal strength (RSS) measurements taken by a cellular handset of the surrounding base stations. RSS location accuracy for different environments is studied as a function of base station separation distance, cell sector density, measurement density, radio propagation environment, and accuracy of measurement. The analytical and experimental results in this thesis serve as a guideline for the accuracy of RSS signature location technology under different conditions. Accurate outdoor to indoor penetration models are proposed and validated for dense urban areas by introducing pseudo-transmitters to simulate the wave-guiding effects in urban canyon environments. A set of location algorithms is developed to improve location accuracy. Furthermore, an algorithm to discriminate between indoor and outdoor users is proposed and validated. The research results demonstrate the feasibility of RSS location techniques to meet the FCCs requirements for E911 accuracy in urban and semi-urban environments. The techniques remain accurate for indoor handsets. The results also suggest that a hybridization of the handset-based GPS method and the RSS signature method may prove to be the most effective solution for locating handsets across a range of environments; including rural, suburban, dense urban, and indoor.

Localisation, which is a mechanism for discovering the spatial relationship between objects, is an area that has received considerable research and development in recent times. A common name given to localisation operations based on the absolute reference frame of Earth is Geolocation. One important example of geolocation research is E-911, where wireless carriers in the United States must provide the location of 911 callers. The operation of E-911 can be based on either a network configuration, or the Global Positioning System (GPS). With the importance of localisation being acknowledged, a review concerning the vulnerability of the Global Navigation Satellite System (GNSS) is provided as background and motivation for this research. With the current vulnerability of GNSS, this dissertation presents the results of a research program undertaken with the objective of developing an electromagnetic localisation technique that can determine the relative position of GPS Radio Frequency Interference (RFI) sources. Intended for operation in a hostile environment, blind and passive localisation methodologies must be incorporated into the developed model. In performing localisation research, a background of current techniques is provided in addition to a review of current electromagnetic propagation models. From the review of propagation models, the Parabolic Equation Model (PEM) was chosen for investigation concerning localisation. The selection of PEM is due to model properties that are required for blind/passive localisation. The localisation system developed in this research program is based on the integration of inverse diffraction propagation (IDP) within the parabolic equation model. The title chosen for the localisation method is Inverse Diffraction Parabolic Equation Localisation System (IDPELS). This thesis presents the simulation and field trial results of IDPELS. Under simulation, the terrain or obstacle profiles were not based on any geodetic datum. Any estimate provided by IDPELS under simulation is therefore a "Localisation" solution. In the field trials however, IDPELS operation is referred to as "Geolocation" as geodetic datum's where used to determine the receiver's position. Under simulation analysis, IDPELS operation was considered to provide good promise as it could simultaneously perform localisation on multiple transmission sources. In each investigated simulation scenario, a display of signals amplitude (dB units) is displayed over the entire region. By determining the field convergence regions, a localisation estimate of IDPELS is provided. By defining the convergence regions as areas having the greatest signal amplitude values (i.e. ≥ 99%), elliptical areas as low as 3.2m² were considered to indicate an excellent localisation capability. With the theoretical validity of IDPELS operation in electromagnetics having been established under simulation, further investigation into the practical feasibility of the IDPELS was performed. The field trials positioned a continuous-wave (CW) transmission source at a known location. By measuring signal phasors along a straight section of road, the geodetic spatial-phase profile was used as the input signal for IDPELS. Road sections used were cross-wise to the transmitter's boresight. Many data sets were recorded, each being made over a sixty second time period. Different regions and ranges where used to continuously measure the spatial-phase profile of the signal with fixed antennas in a moving vehicle. Such a measurement process introduced an analogy with Synthetic Aperture Radar (SAR) processes. In quantitating the accuracy of the IDPELS geolocation estimate in field trials, the linear error of range and cross-range components was analysed. A free-space PEM model was chosen for development of IDPELS and hence, data sets demonstrating properties of a free-space environment were able to be considered suitable for testing of the geolocation method. Data sets demonstrating free-space propagation characteristics were measured at the base of the Mt Lofty ranges in South Australia, where the range and cross-range error are respectively 3.14m, and 0.15m. Such low error values clearly demonstrate the practical feasibility of IDPELS geolocation. With the practical feasibility of IDPELS having been established in this research program, a novel contribution to electromagnetic geolocation methodologies is provided. An important characteristic of any geolocation technique concerns its robustness to operate in a wide variety of possible environments. With continued development of IDPELS, the robustness of this passive/blind geolocation technique can be enhanced. Further assistance with geolocation of multiple transmission sources is also indicated to be available by IDPELS, as shown in the simulation analysis.

Several propagation channel models are designed to verify signal behavior in specific environments for future 5G cellular networks. Hence, with the massive development and demand of indoor wireless systems, a suitable design prior to deployment is required for energy and cost-efficient implementations. These propagation channel models then should be designed to detect weak points in environments and enhance the accuracy of propagation channel modeling in any diverse indoor environment. This work presents the radio propagation analysis of millimeter waves using indoor model-based computational simulations of wireless coverage and material characterization at 28 GHz. Furthermore, the material characterization of common building materials is presented based on excess loss and penetration loss of the propagations in an indoor environment. To perform the analysis, a time static simulator in MATLAB with site-specific maps, deployments and an Indoor 3D propagation model are used to predict path loss in Line of Sight (LoS) and Non-Line of Sight (NLoS) conditions. / Flera spridnings kanal modeller är utformade för att verifiera signal beteende i specifika miljöer för framtida 5Gmobilnät. Med den massiva utvecklingen och efterfrågan på trådlösa inomhussystem krävs därför en lämplig design före driftsättning för energi och kostnadseffektiva implementeringar. Dessa spridnings kanal modeller bör sedan utformas för att upptäcka svaga punkter i miljöer och förbättra noggrannheten i föröknings kanal modellering i alla olika inomhusmiljöer. Detta arbete presenterar radio spridnings analysen av millimetervågor med hjälp av modellbaserade beräknings simuleringar inomhus av trådlös täckning och material karakterisering vid 28 GHz. Dessutom presenteras material karakteriseringen av vanliga byggmaterial baserat på överskottsförlust och penetrations förlust av förökningarna i en inomhusmiljö. För att utföra analysen används en statisk tids simulator i MATLAB med platsspecifika kartor, distributioner och en inomhus 3Dspridnings modell för att förutsäga banförlust i villkor för siktlinje (LoS) och icke-siktlinje (NLoS).

The performance of a mobile satellite communications link is conditioned by the characteristics of the propagation path between a satellite and mobile users. The most important propagation effect in land mobile satellite system is roadside attenuation of the signals due to vegetation or urban structures. System designers should have the most reliable information about the statistics of the propagation channel to build reliable systems that can compensate for bad propagation conditions. In 1998, the Virginia Tech Antenna Group developed a simulator, PROSIM, to simulate a propagation channel in the case of roadside tree attenuation in land mobile satellite systems. This thesis describes some improvements to PROSIM, and the adaptation and validation of PROSIM for Digital Audio Radio Satellite systems operating at S-band frequencies. The performance of the simulator for S-band frequencies was evaluated through a measurement campaign conducted with the XM Radio signals at 2.33 GHz in various propagation environments. Finally, additional results on dual satellite systems and fade correlation are described. / Master of Science