Dissertação (mestrado)—Universidade de Brasília, Faculdade de Tecnologia, Departamento de Engenharia Elétrica, 2015. / Submitted by Albânia Cézar de Melo (albania@bce.unb.br) on 2016-02-16T14:01:50Z No. of bitstreams: 1 2015_RubenOrtegaBlanco.pdf: 78834010 bytes, checksum: 3420abfa3f7efe2d616b97f09d14ff70 (MD5) / Approved for entry into archive by Raquel Viana(raquelviana@bce.unb.br) on 2016-02-19T21:04:29Z (GMT) No. of bitstreams: 1 2015_RubenOrtegaBlanco.pdf: 78834010 bytes, checksum: 3420abfa3f7efe2d616b97f09d14ff70 (MD5) / Made available in DSpace on 2016-02-19T21:04:29Z (GMT). No. of bitstreams: 1 2015_RubenOrtegaBlanco.pdf: 78834010 bytes, checksum: 3420abfa3f7efe2d616b97f09d14ff70 (MD5) / O objetivo do seguinte trabalho é determinar um modelo matemático que permita-nos obter a Relação Sinal Ruído mais Interferência (SNIR do Inglês Signal-to-Noise plus Interference Ratio), a Taxa de Erro de Bits (BER do Inglês Bit Error Rate) de um salto e a Taxa de Erro de Bits fim-a-fim numa rede acústica submarina. Com esse propósito foi desenvolvido um modelo matemático que permite o cálculo destes parâmetros considerando a interferência para o protocolo de acesso ao meio (MAC do Inglês Medium Access Control) ALOHA puro. Também foi necessário desenvolver antes diferentes parâmetros da rede, tais como, distância média do salto, distância média até o nó central, distância média entre os nós, número médio de saltos na rota e desvio médio. Com o uso deste modelo também é possível obter o valor da frequência ótima utilizando uma função de otimização. Comparações entre a Taxa de Erro de Bits de um salto e fim-a fim também forem feitas, para diferentes valores de máximo ângulo de desvio na topologia de rede usada. Estas comparações demonstran quando pode ser mais convenente o uso de um salto ou múltiple-salto. Simulações Monte-Carlo e modelo forem comparados com o propósito de validar os resultados obtidos. Estas comparações demostram a grande similitude entre nosso modelo e as simulações de Monte-Carlo. Além disso foi possível o estudo do comportamento da SNIR e do BER variando importante parâmetros da rede tais como frequência de transmissão, número de nós, raio da esfera e máximo ângulo de desvio. Os resultados obtidos provarem que a SNIR para um salto diminui com o aumento do número de nós e o raio da esfera, mas aumenta com o incremento da potência de transmissão. O comportamento de BER é contrário ao comportamento de SNIR. Também foi possível observar a existência da frequência ótima, onde os melhores valores de SNIR e o BER são obtidos. ______________________________________________________________________________________________ ABSTRACT / The objective of this work is to nd a mathematical model that allow us to obtain the Signal-to-Noise plus Interference Ratio (SNIR), the One-Hop Bit Error Rate (BER) and the End-to-End Bit Error Rate for an Underwater Acoustic Network (UAN). Considering this, it was developed a model that includes the interference as an important impairment and for ALOHA MAC (Medium Access Control) protocol. In addition, it was necessary to obtain before several parameters from the network, such as, average distance of the hop, average distance between nodes, average distance to the central node, average number of hops and average deviation. With this model, it is also possible to nd the optimal value of frequency using an optimization function. It was made comparisons between the One-Hop BER and the End-to-End BER for various values of maximal deviation angle. This comparison shows when it is more adequate to either use one-hop or multi-hop. Finally, we compared numerical and Monte-Carlo simulation results, giving a rst validation to our model. These comparisons show a big similitude between the developed model and the Monte-Carlo simulation. In addition, it was possible to analyse the behaviour of the SNIR and BER by varying important parameters of the network, such as, transmission frequency, number of nodes and sphere radius among others. From the obtained results it was prove that the SNIR decreases with an increase from the number of nodes and the sphere radius, but increases with the transmission power. The End-to-End BER has an contrary behavior with the SNIR.

Les exsudats des plantes sont des substances naturelles complexes. Ils sont utilisés par plusieurs populations d’Afrique et trouvent de nombreuses applications : en thérapeutique, en pharmacie galénique, en parfumerie, etc. Au Congo, une grande majorité de ces substances est encore mal connue. Ce travail de thèse a consisté à inventorier les plantes à exsudats du Congo et à mettre au point une méthodologie d’analyse physico-chimique permettant la caractérisation des exsudats d’arbres d’origine botanique certifiée.La méthodologie d’inventaire adoptée nous a permis de recenser 135 plantes à exsudats appartenant à 87 genres. Il en ressort de cette étude que les exsudats des plantes inventoriées ont multiples usages (glu, insectifuge, allume-feu, antidouleur, etc.).La stratégie mise en place pour leur caractérisation fait appel à plusieurs techniques analytiques (SPIR, SMIR, CPG-SM, GCxGC-MS) et de traitement d’échantillon (Hs-SPME, hydrodistillation, extraction au Soxhlet, silylation, méthylation, pyrolyse). Dans le cadre de ce travail deux exsudats ont été caractérisés. L’exsudat de D. edulis (G. Don) K.J. Lam qui est constitué de monoterpènes hydrocarbonés et de triterpènes, dont les majeurs sont : 3-épi-α et -β-amyrine. L’exsudat de G. demeusei (Harms) J. Léonard dont la fraction volatile est constituée essentiellement de sesquiterpènes et la fraction peu-volatile d’acides diterpéniques de type labdane (acide éperuique, cativique, copalique, ozique, etc.).Pour exploiter l’efficacité de séparation élevée de la GCxGC-MS, nous nous sommes confronté aux problèmes de la non-répétabilité des cartes chromatographiques obtenues en 2D. Nous avons alors mis en place une méthodologie permettant de vérifier l’épaisseur de film et l’homogénéité des dépôts de phases stationnaires afin de sélectionner des colonnes courtes identiques pour la GCxGC. Cependant, les résultats préliminaires ont montré que même avec des colonnes de deuxième dimension identiques, on pouvait avoir des différences au niveau de carte 2D. / Plant exudates are natural complex substances. They are used by several populations in Africa and have many applications: in therapeutics, in galenic pharmacy, in perfumery, etc. In Congo, a great majority of these substances are not yet well known. This thesis work aimed to make an inventory of plant exudates from Congo and to develop a methodology for performing physico-chemical analysis which may enable characterization of exudates from botanical certified trees.Through the methodology that we developed, 135 plant exudates belonging 87 genera were registered. It was observed in this study that these exudates are used for many purposes (glue, insect repellent, making fire, anti-pain, etc).The strategy implemented for characterizing the exudates included several analytical (SNIR, SMIR, GC-MS, GCxGC-MS) and sample processing (Hs-SPME, hydrodistillation, extraction using the Soxhlet, silylation, methylation pyrolyse) techniques. With regard to this work, two exudates were characterized. The exudates from D. edulis (G. Don) K.J. Lam which contained monoterpene hydrocarbons and triterpenes, with the major constituents being: 3-epi-α and -β-amyrine. The exudates from G. Demeusei (Harms) J. Léonard was found to have the sesquiterpenes as major constituents of the volatile fraction and the labdane-type diterpenes (eperuique acide, cativique acid, copalique acid, ozique acid, etc) as major constituents of the poor-volatile fraction.To explore the efficacy of elevated separation of the GCxGC-MS, we faced some problems related to the non-repeatability of chromatographic cards obtained in 2D. Therefore, we put in place a methodology which enables us to verify the film thickness and homogeneity of layers of stationary phases in order to select identical short columns for GCxGC. However, the preliminary results have shown that even using second identical dimension columns, we could observed differences in 2D cards