L'Internet des objets, plus particulièrement les réseaux de capteurs, a attiré beaucoup d'attention ces dernières années. Sa mise en œuvre soulève de nombreuses difficultés, comme la génération d'interférences d'accès multiple (MAI) à caractère impulsif et la consommation d'énergie relativement forte. Les MAI et le bruit thermique doivent être considérés simultanément car ils perturbent fortement les communications. Nous modélisons les MAI et le bruit thermique respectivement par la distribution stable et gaussienne. Nous étudions tout d’abord l’effet des turbo-codes sur le lien direct en utilisant la norme-p comme métrique de décodage. Cette métrique permet une performance de correction d’erreur proche du décodeur optimal. Ensuite nous nous penchons sur les communications coopératives. A l’aide de l’échantillonnage préférentiel, nous estimons les densités de probabilité de la décision statistique du récepteur optimal en présence des bruits stable et gaussien. Cette approche est coûteuse en calcul. Nous proposons donc d’approximer ces densités de probabilité par la distribution gaussienne inverse normale (NIG). Cette solution de calcul est efficace pour approcher le récepteur optimal. Nous montrons également que le récepteur utilisant la norme-p a des performances robustes, quel que soit le type de bruit dominant. A la fin nous combinons les travaux du codage canal et des communications coopératives pour établir une stratégie de codage canal distribué. Basé sur la qualité du lien direct et le niveau de taux d’erreur binaire envisagé, la stratégie d’économie d’énergie peut être mise en place via le choix d’un schéma de codage canal distribué. / Internet of Things brought great interests in recent years for its attractive applications and intelligent structure. However, the implementation of sensor networks still presents important challenges such as the generation of Multiple-Access-Interference (MAI) with impulsive nature and the relatively high energy consumption. Both the MAI and the thermal noise should be considered due to their strong impairments each may cause on the communication quality. We employ the stable and Gaussian distributions to model the MAI and the thermal noise respectively. Firstly we study the performance of turbo codes in the direct link and we propose the p-norm as a decoding metric. This metric allows a considerable error correction performance improvement which is close to the optimal decoder. Then we investigate cooperative communications. The probability densities in the decision statistic of the optimal receiver are estimated using importance sampling approach when both the stable and Gaussian noises are present. Such a method is computationally expensive. Hence we develop an approximation approach based on the Normal Inverse Gaussian (NIG) distribution. This solution is efficient for calculation and is proximate to the optimal receiver. In addition we show that the p-norm receiver has robust performance no matter what kind of noise is dominant. At last we combine the channel coding and cooperative communication works to establish a distributed channel coding strategy. Through some simulation assessments, the energy saving strategy can be realized by choosing an appropriate distributed channel coding scheme based on the direct link quality and target bit error rate.

Le design moderne des systèmes sans fil concerne de plus en plus un déploiement dense d’architectures pour des réseaux sans fil. Par conséquence, l’interférence devient une limitation importante à la performance du système de ces réseaux. Dans plusieurs situations, comme ad hoc sans fil ou transmission de impulsion radio de ultra large bande, les interférences présentent un comportement impulsif. Ce comportement impulsif est souvent mal capté par le modèle Gaussien classique. Avec telles interférences, le récepteur linéaire classique qui est optimal pour un bruit gaussien n’est plus robuste. En plus la plupart des travaux classiques ont été faites sous une hypothèse de variables aléatoires indépendantes et identiquement distribuées interférence. Cependant, une diversité spatiale, temporelle ou fréquentielle peuvent donner des vecteurs un composant dépendant. Dans cette thèse, nous donnons d’abord un modèle général du système et donner plusieurs distributions et modèles de l’interférence impulsive. On évalue ensuite la robustesse de différentes stratégies de récepteur proposé lors des changements de modèle de bruit. Nous proposons différents récepteurs avec différents principaux. Nous donnons également la méthode d’estimation des paramètres et nous illustrons leur performance dans un environnement d’interférence impulsive. Nous proposons également une première approche pour modéliser la dépendance temporelle et spatiale d’interférence. Nous utilisons copule qui permet de séparer les distributions marginales et la structure de dépendance d’interférence. Nous utilisons la famille flexible de skew-t copule et montrer son impact significatif sur la performance d’un récepteur. / Modern wireless system designs are increasingly involving dense deployment architectures for wireless networks. Such a feature makes interference in these networks an important limitation to the system performance. In several situations, such as wireless ad hoc or ultra wide band impulse radio, this interference exhibits an impulsive behavior. Such impulsive behavior is often badly captured by the classical Gaussian model. With such interference, the classical linear receiver, optimal for Gaussian noise, is no longer robust. Most of the conventional works were done under the premise of an assumption of independent and identically distributed interference random variables. However, space, time or frequency diversity can result in vectors with dependent components. In this thesis, we first give and general system model and give several distributions and models, and we compare them in modeling the impulsive interference. We then evaluate the robustness of different receiver strategies proposed when the noise model changes. We propose to classify the different ways to define receivers. We give also the parameter estimation method and we illustrate their performance under an impulsive interference environment. We also propose a first approach to model the time and/or space dependency of the interference samples. We use the framework of copulas that allows separating the marginal distributions and the dependence structure of the interference. We use the flexible family of the skew-t copulas and show that it significantly impacts the performance of a receiver.

Blooms of the red tide dinoflagellate Karenia brevis occur annually on the west Florida shelf. In the late summer/early fall months, background concentrations increase from 103 cells L-1 to excesses of 106 cells L-1. Blooms are most common between Tampa Bay and Charlotte Harbor, and may be maintained for months. The region’s hydrography may play a role in the initiation, maintenance and termination of blooms. The west Florida shelf is depauperate in inorganic nutrients. Inorganic nitrogen and phosphorus rarely exceed the limits of detection, whereas dissolved organic nitrogen is often present at concentrations of 15 to 20 µM. Because K. brevis exhibits the ability to utilize both organic nitrogen and phosphorus, the organic pool may serve as an important nutrient source. The source of nutrients for K. brevis blooms is the focus of much scientific research. Nitrogen is considered to be the limiting nutrient in marine waters and may have several sources. Potential sources of inorganic and organic nitrogen are estuarine outflow, atmospheric deposition, upwelling, dissolved organic nitrogen released from N2 fixing cyanobacteria, diatom blooms, decaying seagrasses, fish or other organic matter. The natural abundance stable isotopic signatures of particulate bloom material (δ15N and δ13C) associated with K. brevis blooms during 1998 to 2001 was analyzed and compared with known isotopic values of potential nutrient sources. Data was analyzed from blooms occurring from 1998 to 2001. Extensive analysis of the 2001 bloom showed that the δ15N of bloom material ranged from 2 0/00 to 5 0/00. δ13C of bloom material ranged from -22 0/00 and -17 0/00. Non-bloom material was considerably more variable in both δ15N and δ13C. δ13C values were higher near shore than offshore during the 2001 bloom, suggesting lower dissolved inorganic carbon levels due to high temperature and/or high biomass. δ15N of bloom material fell within the range of the δ15N values of potential nitrogen sources. It appears that K. brevis utilizes the available nitrogen sources opportunistically, and that isotopically more depleted sources are more important. More enriched sources such as upwelled nitrate or sewage nitrogen can be excluded as significant sources based on the isotopic data.

This thesis addresses some problems that arise in signal processing when the noise is impulsive and follows a heavy tailed distribution. After reviewing several of the more well known heavy- tailed distributions the common problem of which of these hest models the observations is considered. To this end, a test is proposed for the symmetric alpha stable distribution. The test threshold is found using both asymptotic theory and parametric bootstrap resampling. In doing so, some modifications are proposed for Koutrouvelis' estimator of the symmetric alpha stable distributions parameters that improve performance. In electrical systems impulsive noise is generated externally to the receiver while thermal Gaussian noise is generated internally by the receiver electronics, the resultant noise is an additive combination of these two independent sources. A characteristic function domain estimator for the parameters of the resultant distribution is developed for the case when the impulsive noise is modeled by a symmetric alpha stable distribution. Having concentrated on validation and parameter estimation for the noise model, some problems in signal detection and estimation are considered. Detection of the number of sources impinging on an array is an important first. step in many array processing problems for which the development of optimal methods can be complicated even in the Gaussian case. Here, a multiple hypothesis test for the equality of the eigenvalues of the sample array covariance is proposed. / The nonparametric bootstrap is used to estimate the distributions of the test statistics removing the assumption of Gaussianity and offering improved performance for heavy tailed observations. Finally, some robust estimators are proposed for estimating parametric signals in additive noise. These are based on M-estimators but implicitly incorporate an estimate of the noise distribution. enabling the estimator to adapt to the unknown noise distribution. Two estimators are developed, one uses a nonparametric kernel density estimator while the other models the score function of the noise distribution with a linear combination of basis functions.

The present study aimed to determine whether stable isotope techniques can be universally applied to detect landfill leachate contamination in aquatic systems. Results of analysis of ��C in dissolved inorganic carbon ([delta]��C-DIC), deuterium and �⁸O in water ([delta]D-H₂O and [delta]�⁸O-H₂O), and �⁵N of dissolved inorganic nitrogen components ([delta]�⁵N-NH₄⁺ and [delta]�⁵N-NO₃⁻) were presented for leachate, surface, and ground water samples collected from seven landfills located throughout New Zealand between 2003 and 2006. The unique conditions within a landfill lead to measurable fractionations in the isotopic ratios of the products of degradation. Results of isotope and ancillary parameter analyses enabled the discernment of different types of leachate, resulting from different microbial processes within the landfill environment. The isotopic characterisation of leachate enabled improved interpretation of geochemical data from potentially impacted surface and ground waters, and provides useful insight to landfill development for landfill operators. A general isotopic fingerprint delineated by [delta]��C-DIC and [delta]D-H₂O values showed leachate to be isotopically distinct from uncontaminated surface and ground water for samples analysed in the present study. However, not all water samples identified as leachate-impacted via site-specific assessments exhibited isotopic values that overlapped with the general leachate fingerprint. This highlights the need to investigate each site individually, within the context of a possibly global leachate isotope signature. Site-specific investigations revealed the effectiveness of applying [delta]�⁸O-H₂O and [delta]�⁵N-NH₄⁺ or [delta]�⁵N-NO₃⁻, in addition to [delta]��C-DIC and [delta]D-H₂O analyses, to the detection of leachate impact on aquatic systems. Furthermore, ancillary parameters such as alkalinity and ammonium concentration enabled the construction of simple isotope mixing models for an estimate of the quantity of leachate contribution. Results of isotopic investigations of stream biota suggested potential for the development of bio-indicators to monitor leachate influence on aquatic ecosystems in landfill-associated streams. The present study demonstrated the probative power of stable isotope techniques applied to investigations of leachate impact on landfill-associated aquatic systems.

This thesis has developed a suite of compound specific stable isotope tools to monitor landfill leachate and identify the infiltration of leachate to ground water and surface water. These tools have the power to indicate the fractional contribution multiple discrete sources of pollution are making to a single location. This journey began by developing two solid phase extraction (SPE) methods to extract non-polar and polar organic compounds from leachate with minimal fractionation of hydrogen or carbon isotopes. Non-polar compounds were successfully extracted using ENV+ SPE cartridges and polar compounds were successfully extracted using Strata-X SPE cartridges. The isotopic fractionation of non-polar compounds during ENV+ extraction varied significantly (up to 245⁰/₀₀ and 1.8⁰/₀₀ for D and ��C respectively, when eluted with acetonitrile and ethyl acetate, as recommended by manufacturers) but the fractionation of compounds eluted with dichloromethane was negligible (less than instrumental precision). Polar compounds were eluted from Strata-X cartridges with negligible isotopic fractionation using methanol. The direct comparison of SPE and liquid-liquid extraction found SPE to extract slightly more compound from leachate then liquid-liquid extraction (especially for polar compounds) and the isotopic compositions of compounds did not change with extraction methods. These new analytical methods subsequently were used to determine the isotopic compositions of organic compounds dissolved in leachates from three New Zealand landfills. The molecular and isotopic signature of leachate varied significantly between landfills, indicating the isotopic fingerprint of organic compounds in leachate is unsuitable as a universal tracer of leachate. However, compounds such as terpien-4-ol, methylethylbenzene and juvabione maintained their isotopic composition over short geographical distance-indicating their potential as site-specific tracers of leachate. Organic compounds analysed on a transect across the landfill boundary indicated polar compounds were more mobile than semi-volatile compounds and possessed a more conservative isotopic composition. However, hexadecanoic acid extracted from leachate and ground water was highly depleted in ��C (-72 ⁰/₀₀ to -40⁰/₀₀), indicative of methanogenic and sulfate reducing bacteria. These bacteria only live in highly reducing environments such as leachate; therefore their presence in the pristine environment can potentially indicate the release of leachate from the landfill. The final experiments traced the uptake and utilisation of leachate by periphyton. The isotopic composition of bulk periphyton, fatty acids and phytol indicated that microbial assimilation and utilisation of nutrients is a complex process. Fatty acid biomarkers for green algae and diatoms showed signs of leachate derived nutrients, however the availability of nutrients (carbon, nitrogen, water and light) caused significant changes in metabolic processes and isotopic compositions. Under slow growing conditions, the [delta]��C composition of periphyton became enriched in ��C as solar irradiation levels decreased (including shading by detritus and periphyton), while the [delta]D composition of fatty acid was controlled by the internal recycling of hydrogen. This study indicated the power of compound specific isotope analysis as a tool to detect the release of landfill leachate from a landfill, especially at locations with multiple potential sources of contaminants, and provides a sound platform for future research.

On étudie la norme stable sur l'homologie des variétés riemanniennes, plus spécialement dans le cas où la variété est une surface orientable de genre supérieur à 1. On établit le lien entre norme stable et fonction beta de Mather, puis on montre des résultats de non-stricte convexité et non-différentiabilité de la norme stable. Ensuite on compare la norme stable et la norme L2.

<p>In this project I have tried to obtain stable transformants of <em>Drosophila</em> melanogaster flies using electroporation. I have completed approximately 200 tests using different DNA concentrations, voltages and cuvettes, including a novel Petri dish cuvette which I developed and manufactured myself. I also developed new and more efficient procedures of egg collection and egg dechorionation. Although I was not  successful in obtaining true stable transformants, control experiments indicate that electroporation of DNA into embryos could be accomplished under the conditions used. The lack of stable transformants was probably due to failure of the electroporated DNA to integrate into the host genome. The reasons for why the DNA did not integrate was not further investigated in this study.</p>

An increasing number of parameter estimation tasks involve the use of at least two information sources, one complete but limited, the other abundant but incomplete. Standard algorithms such as EM (or em) used in this context are unfortunately not stable in the sense that they can lead to a dramatic loss of accuracy with the inclusion of incomplete observations. We provide a more controlled solution to this problem through differential equations that govern the evolution of locally optimal solutions (fixed points) as a function of the source weighting. This approach permits us to explicitly identify any critical (bifurcation) points leading to choices unsupported by the available complete data. The approach readily applies to any graphical model in O(n^3) time where n is the number of parameters. We use the naive Bayes model to illustrate these ideas and demonstrate the effectiveness of our approach in the context of text classification problems.

The boreal chorus frog (Pseudacris maculata) is a widespread species but we know little of its ecology. I examined the nature and existence of competitive mechanisms operating between larvae of the boreal chorus frog and wood frog (Lithobates sylvaticus) from field, mesocosm and laboratory venues spanning nutrient concentrations. I assessed larval performance and diet of tadpoles at natural ponds by measuring tadpole growth and size at metamorphosis, and stable isotope ratios for carbon and nitrogen in tadpole tissue to examine if patterns were consistent with the operation of interspecific competition. In mesocosms I measured chorus frog performance in relation to wood frog presence and nutrient enrichment to confirm the occurrence of competition and examine whether nutrient conditions typical of agriculture ponds impact tadpole performance. In the field I compared larval performance and relative abundance between agricultural ponds and those in Elk Island National Park, to examine whether habitat features surrounding ponds in farmlands reduces the abundance of tadpoles and whether tadpole performance results in reduced abundances. Lastly, I examined whether chemical interference by wood frogs occurs by raising chorus frog tadpoles with caged wood frog tadpoles and/or their feces in the laboratory. Chorus frog performance was reduced by presence and abundance of wood frog tadpoles. Resource partitioning in natural ponds and overlap in mesocosms, based on stable isotopic analysis, suggest that resource competition occurs. In mesocosms chorus frog performance was reduced by wood frog tadpoles in fertilized treatments and nutrient conditions at agricultural sites are not in themselves detrimental to these anurans. Performance of chorus frog tadpoles in agricultural ponds was unaffected, whereas wood frogs were larger at metamorphosis in crop ponds. Reduced tadpole abundances of both species at these ponds may be related to habitat features or conditions in croplands. The existence of chemical interference in the absence of physical interaction was confirmed, as chorus frogs exposed to wood frog tadpoles and/or feces had reduced growth and were smaller at metamorphosis which could reduce terrestrial survival and future reproduction. My research contributes to our knowledge on boreal chorus frog ecology and our general understanding of competition between larval anurans. / Ecology