In this thesis, the neural network, a powerful tool which has demonstrated its ability in many fields, is studied for the indoor localization system and channel prediction system. This thesis first proposes a received signal strength indicator (RSSI) fingerprinting-based indoor positioning system for the widely deployed WiFi environment, using deep neural networks (DNN). To reduce the computing time as well as improve the estimation accuracy, a two-step scheme is designed, employing a classification network for clustering and several regression networks for final location prediction. A new fingerprinting, which utilizes the similarity in RSSI readings of the nearby reference points (RPs) is also proposed. Real-time tests demonstrate that the proposed algorithm achieves an average distance error of 43.5 inches. Then this thesis extends the ability of the neural network to the physical layer communications by introducing a recurrent neural network (RNN) based approach for real-time channel prediction which uses the recent history channel state information (CSI) estimation for online training before prediction, to adapt to the continuously changing channel to gain a more accurate CSI prediction compared to the other conventional methods. Furthermore, the proposed method needs no additional knowledge, neither the internal properties of the channel itself nor the external features that affect the channel propagation. The proposed approach outperforms the other methods in a changing environment in the simulation test, validating it a promising method for channel prediction in wireless communications. / Graduate

Este estudo buscou compreender o canal de marketing usado para distribuição de servidores de microinformática do tipo x86 que, segundo institutos de pesquisas como o IDC (2009), responderão pela maioria absoluta de todos os servidores vendidos no país. Dentro da categoria de servidores x86, foram explorados os que estão em fases de crescimento e maturidade no ciclo de vida, devido aos potenciais de venda e de agregação de serviços ao cliente final. Para tanto, foram entrevistados doze membros deste canal, incluindo o fabricante dos produtos, três atacadistas, quatro varejistas e quatro clientes finais. Estes foram questionados sobre suas funções no canal de marketing analisado, avaliação de capacitação técnica para venda e prestação de serviços ao cliente, além dos papéis dos clientes e suas expectativas quando ao serviço ao cliente prestado. Percebeu-se que deficiência na capacitação e falta de estruturação nos varejistas pode ser um fator crítico para o canal de marketing, podendo comprometer a imagem dos produtos vendidos, o entendimento das necessidades e a prestação de serviços ao cliente e, de certa forma, o resultado de todos os outros membros do canal. / The intent of this study was to understand the marketing channel used to distribute x86 servers which, according to research institutes such as IDC (2009), will account for the majority of the servers sold in Brazil. Within the x86 category only products in growth or maturing lifecycle were considered in the investigation. These were selected due to potential of sales and customer services aggregation to the final purchasing customers. To accomplish that, twelve members of this channel were interviewed, including the product manufacturer, three wholesalers, four retailers and four final customers. They were asked about their function in the studied marketing channel, about their technical capability to sell and to serve customers, in addition to understanding the customer roles and their expectations on the performed service. The results found that reseller deficiency in technical skills and poor structure definition may be a critical factor for the marketing channel. This possibly compromises the image of sold products, the understanding of customer needs, customer services delivered and, in a certain way, the results for the other channel members.

The use of fatty acid (FA) analysis has become an important tool in recent years to investigate a broad spectrum of questions in fisheries and aquatic ecology. One of these applications has been the use of FA profiles as biomarkers for indicating diet and recent feeding history of fishes and other aquatic consumers. Differences in forage FA profiles and spatial differences in food web structure within aquatic systems may lead to corresponding spatial patterns of FA profiles in consumers, thereby facilitating the potential applicability of FA biomarkers as indicators of habitat use and origin for mobile consumers such as fishes. However, little information is available regarding differences in FA profiles among habitat types in river-floodplain ecosystems and the time-scale over which fish FA profiles change when fish move between habitats that differ in the FA profiles of prey resources. The objectives of this study were to test whether the tissue FA profiles of channel catfish (Ictalurus punctatus) differed among three reaches of the lower Kaskaskia River and its floodplain lakes, to determine the time-course of tissue FA profile turnover when fish are relocated to a new habitat type, and to compare FA profiles among muscle, liver, and adipose fin tissues in channel catfish. Fatty acid profiles of channel catfish were significantly different among sites, especially between upper and lower river sites, and between river channel and oxbow lake sites, suggesting differing energy sources among habitats and river reaches in the Kaskaskia system. More specifically, there was a significant increase in the essential FAs, 18:2n-6 and 18:3n-3, in channel catfish with increasing distance downstream, which could reflect an increase in river-floodplain connectivity at downstream sites. Channel catfish transplanted from the Kaskaskia River to ponds at Southern Illinois University Carbondale showed a significant linear decrease in n-3 and medium-chain polyunsaturated FAs (MC-PUFA), and a significant linear increase in monounsaturates (MUFA) with longer residence time in a lentic environment. Liver tissue exhibited the fastest turnover rate of the three tissue types (< 2 weeks), with muscle and adipose fin tissue both displaying similar, longer turnover times (approximately 10 weeks). Results of this study support the use of FA profiles as indicators of energy sources for fishes in large river-floodplain ecosystems, including the potential for indentifying habitat-specific (river channel vs. floodplain lake) or river reach-specific energy sources. Additionally, results of this study provide a timeline for FA turnover in channel catfish tissues when fish move among habitats with distinct FA signatures, which is crucial for detecting temporal shifts in use of habitat-specific energy sources by channel catfish. Results also suggest that adipose fin tissue samples may be a less invasive alternative to muscle tissue for analysis of FA profiles in channel catfish. Improved understanding of which habitats or river reaches provide trophic support for riverine fishes through use of FA profile analysis has potentially important implications for habitat conservation and rehabilitation in river-floodplain ecosystem.

Metals are ubiquitous and naturally occurring; however, anthropogenic activities have elevated metal concentrations in river sediments above what would be expected naturally. One of the primary anthropogenic sources of metals in freshwater is contamination by mine tailings, a toxic slurry of waste rock and chemicals left over after ore minerals or coal extraction. Mining waste is most often stored in retention ponds, which sometimes leak or fail. One of the worst mine tailings disasters on record occurred on the Tisza River of Central Europe in the winter of 2000, releasing over 240,000 tons of mine tailings laced with metals into the natural environment. Elevated metals in freshwater can negatively affect fish and other aquatic organisms, raising concerns as to the long-term ecological consequences of the spill. Therefore, the objectives of this dissertation were to describe the ecological implications of metals pollution from mining on freshwater systems through an empirical case study of the Tisza River combined with controlled laboratory experiments of chronic cadmium exposure on a model species, channel catfish (Ictalurus punctatus). Specific research questions discussed in the following chapters include: (1.) What is the current distribution of metals in the lower Tisza River Basin and are metals transported from the main channel to floodplain habitats?; (2.) Is there evidence for bioaccumulation and/or trophic transfer of metals to fish in the Tisza?; (3.) What are angler’s perceptions and fish consumption risks in the lower Tisza river basin?; and (4.) What are the effects of sub-lethal exposure to cadmium on growth, development, metabolism, and stress response in a model fish species? The case study of the Tisza River Basin was conducted during early summer in 2013 and 2014. Water, sediment, and fish were collected from the Tisza River as well oxbow lakes along the lower basin. Samples were screened for metals via GF-AAS following standard methods. Additionally, we conducted interviews at each sampling location to determine fish consumption habits. Controlled laboratory exposures were conducted during summer of 2015. Eggs of channel catfish (Ictalurus punctatus) were fertilized in treatment water and raised to 6 month old fingerlings. Treatments of cadmium included concentrations of 0.5 µg L-1 (control), 2 L-1 (low) and 6 L-1 (high), with endpoints of growth, development, cellular stress, metabolism, and general stress response. Results indicate that lakes with the greatest connectivity trended toward elevated metals; however, chlorophyll a concentrations decreased suggesting dilution of nutrients from surrounding agriculture. River connectivity therefore increases ecosystem health of floodplain lakes by ameliorating eutrophication, but as a trade-off with potential contamination of metals. This may have implications for management strategies in the basin as fish from the oxbow lakes also trended toward higher concentrations of metals compared to the river main stem. Although we did not detect any indication that metals are biomagnifying through the fish community, fish are clearly taking up contaminants from the water and sediment. Additionally, although fish fillets generally fell below human consumption guidelines, surveys collected at field sites indicate disparities between health risks and perceptions of those risks. For example, many people believe that fish from the oxbow lakes present a lower risk for metals than the river main stem, converse to our findings. Exposure of channel catfish to cadmium concentrations similar to those observed in the Tisza Basin indicate that negative physiological effects, such as altered carbohydrate metabolism and subsequently growth, can occur in fish at muscle concentrations below consumption guidelines. Although cadmium in Tisza fish fillets were lower than expected, metals may still be problematic for the basin as they may disproportionately affect early life stages of fish. This could skew life histories of exposed populations compared to unaffected fish, potentially reducing growth rates, size at maturity, reproductive output, and lifespan; indicating the need for more paired field and laboratory assessments of chronic metals exposure.

xiii, 229 p. : ill. A print copy of this title is available from the UO Libraries, under the call number: KNIGHT PN4888.R33 S26 2008 / This study looks at the status of local radio news in the United States in light of changes in policy, economics, production and distribution technology and the dynamic media environment. It examines how differences in ownership relate to the amount of news programming offered on local stations, how those stations are staffed and the working conditions for today's radio journalists. Two areas of communication theory provide the basis for the study, Political Economy of Communication and Localism. Both offer excellent perspectives for studying the radio broadcasting industry and the people who work in it. Political economy allows the study to look closely at the impact of ownership in our capitalist society, how government regulates ownership and programming, how those factors affect the working conditions for journalists and how they ultimately impact the public interest. Political economy is a holistic approach that also calls upon us to consider a moral philosophy and make recommendations for the good of society. Localism is a long-held policy objective of the Federal Communications Commission that has been a part of the regulatory process relating to ownership and programming of news and public affairs throughout the existence of radio in the United States. Using a triangulation of both quantitative and qualitative methods, the study documents the news operations of four different types of ownership structures within a single radio market, Salt Lake City, Utah. The primary quantitative method used content analysis to examine a sample of 255 hours of radio programming across the ownership groups. Qualitative methods of in-depth interviews and observation were used to examine how the stations were staffed, the working conditions for local journalists and how the news programming is produced. The study found the overall amount of local radio news programming to be low, with locally owned stations generally producing more news then those with large, outside corporate ownership. It also found working conditions to vary greatly among ownership groups. Local owners tended to be much more supportive of local journalists and provide better conditions for the production and programming of local radio news. / Adviser: Alan G. Stavitsky

This paper presents co-resident watermarking, a traffic analysis attack for cloud environments that allows a malicious co-resident virtual machine to inject a watermark signature into the network flow of a target instance. This watermark can be used to exfiltrate co-residency data, compromising isolation assurances. While previous work depends on virtual hypervisor resource management, our approach is difficult to defend without costly underutilization of the physical machine. We evaluate co-resident watermarking under many configurations, from a local lab environment to production cloud environments. We demonstrate the ability to initiate a covert channel of 4 bits per second, and we can confirm co-residency with a target VM instance in less than 10 seconds. We also show that passive load measurement of the target and behavior profiling is possible. Our investigation demonstrates the need for the careful design of hardware to be used in the cloud. This thesis includes unpublished co-authored material.

The effect of climate on the fluvial system has long been investigated due the significant impact it can have on a river’s hydrological regime and fluvial processes. In recent years this interest has increased as global changes in climate are expected to bring more frequent high magnitude flood events globally and to North West Europe in particular. Despite the knowledge that the frequency and magnitude of floods is to increase, less is known about the geomorphological implications of this for river channels and where channel instability is likely to occur at both the river network and national scale. This is certainly the case in Scotland where increased flooding is expected and large floods have been abundant over the last two decades. To manage Scottish river catchments effectively in the future, in terms of hazard mitigation and nature conservation, river managers need to be able to predict not only how climate will impact flood magnitude and frequency in Scotland but the effect these changes will have on the internal dynamics of river channels in terms of erosion, sediment transport and deposition, and morphological dynamics. Such knowledge will ensure adequate measures are implemented to reduce fluvial risks to humans and to maintain and preserve valuable river habitats and linked species. In this thesis, several novel methods incorporating field, laboratory and GIS-based analysis, have been investigated as a means of predicting how climate change will affect channel stability in Scottish rivers and the implications of this for river management and river ecology. This includes (i) analysing the potential change in the frequency of geomorphologically-active flood flows with climate change; (ii) the use of stream power thresholds to predict changes in channel stability on a national scale with climate change; and (iii) using a Digital River Network developed using geospatial data to predict changes in the rate of bedload transfer and channel stability with climate change. Studies were undertaken on 13 different rivers across Scotland from north to south and east to west. vii As a case study of ecological implications, the thesis also examines how changes in habitat and stability of freshwater pearl mussels (Margaritifera margaritifera) may be altered by increased flooding. Predictions of the frequency of geomorphic activity, channel stability, rate of bedload transfer, and the stability of freshwater pearl mussel habitat with climate change are discussed along with the methods used to obtain these outcomes. The results all suggest an increase in the frequency and rate at which bedload is transferred through the river system and an increased frequency of flood flows resulting in greater channel instability. Morphological responses vary spatially with some river reaches experiencing greater increased erosion and transport potential than others. Climate change effects on the freshwater pearl mussel are: increased occasions of disturbance and transport downstream and the importance of specific populations in more stable environments for ensuring population recovery post flooding is highlighted. It is hoped that the methodologies developed for predicting changes in channel stability with climate change will provide useful screening tools to regulatory agencies which can be developed further to assist management decisions in the future which aim to reduce fluvial hazards and maintain good quality river environments for the species that inhabit it. The approaches used in this study allow for the identification of areas at high risk of morphological and ecological change, and the pro-active planning and management of sediment-related river management issues and nature conservation.

La performance de démodulation des signaux GNSS existants, GPS L1 C/A, L2C ou L5, est satisfaisante en environnements ouverts où le C/N0 disponible est assez élevé. Cependant, en milieu urbain, le niveau de C/N0 du signal reçu est souvent très bas et est affecté de variations rapides qui peuvent nuire la démodulation des messages GNSS. Donc, car les applications du marché de masse sont appelées à être déployées dans ces environnements, il est nécessaire d'étudier et de chercher des méthodes de démodulation/décodage qui améliorent la performance de démodulation des messages GNSS dans ces environnements. Il est aussi nécessaire de considérer les nouveaux signaux GPS L1C et GALILEO E1. Ces signaux doivent fournir un service de positionnement par satellite dans tout type d'environnement, et spécifiquement en milieu urbain. Ainsi, cette thèse analyse aussi les performances de démodulation des nouveaux signaux GNSS tels que définis dans les documents publics actuels. De plus, de nouvelles structures de message GALILEO E1 sont proposées et analysées afin d'optimiser la performance de démodulation ainsi que la quantité d'information diffusée. En conséquence, le but principal de cette thèse est d'analyser et améliorer la performance de démodulation des signaux GNSS ouverts au public, spécifiquement en milieu urbain, et de proposer de nouvelles structures de messages de navigation pour GALILEO E1. La structure détaillée des chapitres de cette thèse est donnée ci-après. En premier lieu, le sujet de cette thèse est introduit, ses contributions originales sont mises en avant, et le plan du rapport est présenté. Dans le 2ième chapitre, la thèse décrit la structure actuelle des signaux GNSS analysés, en se concentrant sur la structure du message de navigation, les codages canal implantés et leurs techniques de décodage. Dans le 3ième chapitre, deux types de modèles de canal de propagation sont présentés pour deux différents types de scénarios. D'un côté, un canal AWGN est choisi pour modéliser les environnements ouverts. De l'autre côté, le modèle mathématique de Perez-Fontan d'un canal mobile est choisi pour représenter les environnements urbains et indoor. Dans le 4ième chapitre, une tentative pour effectuer une prédiction binaire d'une partie du message de navigation GPS L1 C/A est présentée. La prédiction est essayée en utilisant les almanachs GPS L1 C/A, grâce à un programme de prédiction à long terme fourni par TAS-F, et des méthodes de traitement du signal: estimation spectrale, méthode de PRONY et réseau de neurones. Dans le 5ème chapitre, des améliorations à la performance de démodulation du message de GPS L2C et L5 sont apportées en utilisant leur codage canal de manière non traditionnelle. Deux méthodes sont analysées. La première méthode consiste à combiner les codages canal internes et externes du message afin de corriger davantage de mots reçus. La deuxième méthode consiste à utiliser les probabilités des données d'éphémérides afin d'améliorer le décodage traditionnel de Viterbi. Dans le 6ième chapitre, la performance de démodulation des messages de GPS L1C et du Open Service GALILEO E1 est analysée dans différents environnements. D'abord, une étude de la structure de ces deux signaux est présentée pour déterminer le C/N0 du signal utile reçu dans un canal AWGN. Puis, la performance de démodulation de ces signaux est analysée grâce à des simulations dans différents environnements, avec un récepteur se déplaçant à différentes vitesses et avec différentes techniques d'estimation de la phase porteuse du signal. / The demodulation performance achieved by any of the existing GPS signals, L1 C/A, L2C or L5, is satisfactory in open environments where the available C/N0 is quite high. However, in indoor/urban environments, the C/N0 level of the received signal is often very low and suffers fast variations which can further affect the GNSS messages demodulation. Therefore, since the mass-market applications being designed nowadays are aimed at these environments, it is necessary to study and to search alternative demodulation/decoding methods which improve the GNSS messages demodulation performance in these environments. Moreover, new GNSS signals recently developed, such as GPS L1C and GALILEO E1, must also be considered. These signals aim at providing satellite navigation positioning service in any kind of environment, giving special attention to indoor and urban environments. Therefore, the demodulation performances of the new GNSS signals as they are defined in the current public documents is also analysed. Moreover, new GALILEO E1 message structures are proposed and analysed in order to optimize the demodulation performance as well as the quantity of broadcasted information. Therefore, the main goal of this dissertation is to analyse and to improve the demodulation performance of the current open GNSS signals, specifically in indoor and urban environments, and to propose new navigation message structures for GALILEO E1. A detailed structure of this dissertation sections is given next. First, the subject of this thesis is introduced, original contributions are highlighted, and the outline of the report is presented. Second, this dissertation begins by a description of the current structure of the different analysed GNSS signals, paying special attention to the navigation message structure, implemented channel code and their decoding techniques. In the third section, two types of transmission channel models are presented for two different types of environments. On one hand, an AWGN channel is used to model the signal transmission in an open environments. On the other hand, the choice of a specific mobile channel, the Perez-Fontan channel model, is chosen to model the signal transmission in an urban environment. In the fourth section, a tentative to make a binary prediction of the broadcasted satellite ephemeris of the GPS L1 C/A navigation message is presented. The prediction is attempted using the GPS L1 C/A almanacs data, a long term orbital prediction program provided by TAS-F, and some signal processing methods: spectral estimation, the PRONY method, and a neural network. In the fifth section, improvements to the GPS L2C and GPS L5 navigation message demodulation performance are brought by using their channel codes in a non-traditional way. Two methods are inspected. The first method consists in sharing information between the message inner and outer channel codes in order to correct more received words. The second method consists in using the ephemeris data probabilities in order to improve the traditional Viterbi decoding. In the sixth section, the GPS L1C and GALILEO E1 Open Service demodulation performance is analysed in different environments. First, a brief study of the structure of both signals to determine the received C/N0 in an AWGN channel is presented. Second, their demodulation performance is analysed through simulations in different environments, with different receiver speeds and signal carrier phase estimation techniques.

Ce travail de thèse porte sur le développement de méthodes de réduction du canal de multitrajets, donnant une importance particulière à la préservation des performances en poursuite du signal par les systèmes de navigation par satellite. Le but est de rendre les modèles de canal mobile urbain applicables au test réaliste de nouveaux récepteurs GNSS, en réduisant drastiquement le nombre d’échos synthétisés par le simulateur de canal d’origine (de plusieurs milliers à moins de 10) tout en conservant son impact sur l’erreur de pseudo-distance. Trois types de méthodes de réduction ont été envisagés afin de couvrir l’ensemble des possibilités de réduction. En première approche, une méthode d’agrégation a été développée, par le clustering des multitrajets selon les dimensions du retard et du Doppler pondéré par leur puissance. Bien que cette méthode permette une bonne préservation des caractéristiques large-bande du canal d’origine, une deuxième approche a été considérée, visant à optimiser les paramètres des échos du canal réduit (retard, Doppler, phase et amplitude) pour minimiser la différence entre la fonction de corrélation du canal d’origine et celle du canal réduit. Bien que cette approche donne de bonnes performances en termes de préservation de l’erreur de pseudodistance, elle conduit à une instabilité des échos du canal réduit. C’est dans cette optique qu’une approche statistique a été développée, considérant que le retard des multitrajets évolue selon une chaîne de Markov du premier ordre. Ces méthodes ont été comparées selon le critère de la préservation de l’erreur du discriminateur en boucle ouverte. Dans certaines conditions portant sur le nombre de multitrajets du canal réduit et l’élévation du satellite émetteur, il apparait que la méthode de clustering implémentée en première approche égale les performances de la méthode paramétrique, la méthode statistique donnant systématiquement des performances inférieures. L’invariance de cette hiérarchie lors de l’ajout de boucles de poursuite à la simulation ou le test de divers modèles de signaux (modulation et bande RF) permet d’étendre ces conclusions. Etant donné l’avantage significatif des méthodes de clustering sur les méthodes paramétriques en termes de temps de calcul, c’est l’usage de cette méthode qui est préconisé pour répondre à la problématique de réduction du canal de multitrajets. En particulier, la méthode de clustering pondéré développée dans cette thèse offre la possibilité de reproduire les paramètres largebande d’un canal composé de milliers de multitrajets avec moins de 10 échos. De plus, elle offre un compromis intéressant entre temps de calcul et préservation de l’erreur de pseudo-distance des systèmes GNSS, approchant les performances de méthodes d’optimisation paramétrique, voire les atteignant dans certaines conditions, avec un temps de calcul proche du temps réel. / This PhD work relates the development of multipath channel reduction methods, putting the emphasis on the preservation of the signal tracking performances of satellite navigation systems. It aims at adapting the land mobile channel models to the realistic testing of GNSS receivers, by reducing drastically the number of multipaths of the original channel model (from thousands to less than 10 typically) all by conserving the pseudo-range error. Three types of methods have been investigated in order to cover all different possibilities of reduction processes. As a first approach, a multipath aggregation method has been considered, through the clustering of the multipaths according to the delay and Doppler dimensions, weighted by their power. Even if this method allows a good preservation of the delay and Doppler characteristics of the original channel, a second approach has been investigated, oriented toward the optimization of the parameters of the reduced channel (delay, Doppler shift, phase and amplitude) in order to minimize the difference between the original channel correlation function and the reduced one. Even if this approach gives good performances in terms of pseudorange error preservation, it leads to the instability of the reduced channel echoes. Therefore, a statistical method has been implemented, considering that the delays of the reduced channel evolve according to a first order Markov process. These three approaches have been compared through the preservation of the discriminator open loop error. In particular conditions concerning the number of echoes in the reduced channel or the elevation of the emitter satellite, the clustering technique implemented in first approach appears to equal the performances of the parametric method, the statistical approach giving systematically the worst performances. The invariance of this hierarchy as adding tracking loops to the simulation or changing the signal model (modulation and RF bandwidth) allows extending the conclusion. Being given the significant advantage of the clustering on the parametric method in terms of computation time, the use of this method is preconized to address the channel reduction problem. In particular, the weighted clustering technique developed in this thesis offers the possibility to reproduce the wide-band characteristics of a channel model composed of thousands of multipaths with less than 10 echoes. Moreover, it constitutes an interesting tradeoff between computational effort and GNSS systems pseudo-range error conservation, approaching the performances of parametric methods, and even overcoming them in some conditions, with a computation time close to the real time.

A economia mundial acompanha o avanço da ciência e da tecnologia, enquanto o preço do trabalho e dos recursos materiais tornam-se cada vez mais caros, forçando empresas a buscarem reduções em seus custos para, consequentemente, atingirem um maior nível de rentabilidade. Nesse contexto, as vending machines, além de serem caracterizadas pelo intenso grau de inovação tecnológica, possuem uma concepção importante no sentido de oferecerem menores custos de instalação e de mão de obra. No Brasil, a utilização de vending machines vem crescendo de maneira acelerada desde a década passada, especialmente junto a empresas fabricantes de produtos de conveniência, e essas máquinas tornam-se cada vez mais importantes, consolidando esse inovador sistema de distribuição. Diante disso, a presente pesquisa buscou compreender qual o papel das vending machines para empresas fabricantes de produtos de conveniência no Brasil enquanto membros do canal de distribuição. O método de pesquisa utilizado foi o estudo multicaso, sendo quatro o número de unidades de análise que concordaram em participar da pesquisa. Os resultados sugerem que os principais aspectos que compõem o papel das vending machines estão a relacionados à divulgação de marcas e à fidelização dos clientes, bem como à realização de novas vendas e à comercialização de produtos de maior valor agregado. / The global economy follows the advance of science and technology, while the price of labor and material resources become increasingly expensive, forcing companies to seek reductions in their costs, so, consequently, achieve higher levels of profitability. In this context, vending machines, besides being characterized by intense degree of technological innovation, have an important conception in order to offer lower installation costs and labor. In Brazil, the use of vending machines has been growing at an accelerated pace since the past decade, especially with the manufacturers of convenience products, and these machines become increasingly important, consolidating this innovative distribution system. Therefore, this study aimed to understand the role of vending machines for manufacturers of convenience products in Brazil as members of distribution channel. The method used was the multi case study, and four was the number of units of analysis which have agreed to participate in the research. The results suggest that the main aspects which make up the role of vending machines are related to the disclosure of brands and customer loyalty, as well as the implementation of new sales and the transaction of products with higher added value.