Flexible numerology of the physical layer has been introduced in the latest release of 5G new radio (NR) and the baseline waveform generation is chosen to be cyclic-prefix based orthogonal frequency division multiplexing (CP-OFDM). Thanks to the narrow subcarrier spacing and low complexity one tap equalization (EQ) of OFDM, it suits well to time-dispersive channels. For the upcoming 5G and beyond use-case scenarios, it is foreseen that the users might experience high mobility conditions. While the frame structure of the 5G NR is designed for long coherence times, the synchronization and channel estimation (CE) procedures are not fully and reliably covered for diverse applications. The research on alternative multi-carrier waveforms has brought up valuable results in terms of spectral efficiency, applications coexistence and flexibility. Nevertheless, the receiver design becomes more challenging for multiple-input multiple-output (MIMO) non-orthogonal multi-carriers because the receiver must deal with multiple dimensions of interference. This thesis aims to deliver accurate pilot-aided estimations of the wireless channel for coherent detection. Considering a MIMO non-orthogonal multi-carrier, e.g. generalized frequency division multiplexing (GFDM), we initially derive the classical and Bayesian estimators for rich multi-path fading channels, where we theoretically assess the choice of pilot design. Moreover, the well time- and frequency-localization of the pilots in non-orthogonal multi-carriers allows to reuse their energy from cyclic-prefix (CP). Taking advantage of this feature, we derive an iterative approach for joint CE and EQ of MIMO systems. Furthermore, exploiting the block-circularity of GFDM, we comprehensively analyze the complexity aspects, and propose a solution for low complexity implementation. Assuming very high mobility use-cases where the channel varies within the symbol duration, further considerations, particularly the channel coherence time must be taken into account. A promising candidate that is fully independent of the multi-carrier choice is unique word (UW) transmission, where the CP of random nature is replaced by a deterministic sequence. This feature, allows per-block synchronization and channel estimation for robust transmission over extremely doubly-dispersive channels. In this thesis, we propose a novel approach to extend the UW-based physical layer design to MIMO systems and we provide an in-depth study of their out-of-band emission, synchronization, CE and EQ procedures. Via theoretical derivations and simulation results, and comparisons with respect to the state-of-the-art CP-OFDM systems, we show that the proposed UW-based frame design facilitates robust transmission over extremely doubly-dispersive channels.:1 Introduction 1 1.1 Multi-Carrier Waveforms 1 1.2 MIMO Systems 3 1.3 Contributions and Thesis Structure 4 1.4 Notations 6 2 State-of-the-art and Fundamentals 9 2.1 Linear Systems and Problem Statement 9 2.2 GFDM Modulation 11 2.3 MIMO Wireless Channel 12 2.4 Classical and Bayesian Channel Estimation in MIMO OFDM Systems 15 2.5 UW-Based Transmission in SISO Systems 17 2.6 Summary 19 3 Channel Estimation for MIMO Non-Orthogonal Waveforms 21 3.1 Classical and Bayesian Channel Estimation in MIMO GFDM Systems 22 3.1.1 MIMO LS Channel Estimation 23 3.1.2 MIMO LMMSE Channel Estimation 24 3.1.3 Simulation Results 25 3.2 Basic Pilot Designs for GFDM Channel Estimation 29 3.2.1 LS/HM Channel Estimation 31 3.2.2 LMMSE Channel Estimation for GFDM 32 3.2.3 Error Characterization 33 3.2.4 Simulation Results 36 3.3 Interference-Free Pilot Insertion for MIMO GFDM Channel Estimation 39 3.3.1 Interference-Free Pilot Insertion 39 3.3.2 Pilot Observation 40 3.3.3 Complexity 41 3.3.4 Simulation Results 41 3.4 Bayesian Pilot- and CP-aided Channel Estimation in MIMO NonOrthogonal Multi-Carriers 45 3.4.1 Review on System Model 46 3.4.2 Single-Input-Single-Output Systems 47 3.4.3 Extension to MIMO 50 3.4.4 Application to GFDM 51 3.4.5 Joint Channel Estimation and Equalization via LMMSE Parallel Interference Cancellation 57 3.4.6 Complexity Analysis 61 3.4.7 Simulation Results 61 3.5 Pilot- and CP-aided Channel Estimation in Time-Varying Scenarios 67 3.5.1 Adaptive Filtering based on Wiener-Hopf Approac 68 3.5.2 Simulation Results 69 3.6 Summary 72 4 Design of UW-Based Transmission for MIMO Multi-Carriers 73 4.1 Frame Design, Efficiency and Overhead Analysis 74 4.1.1 Illustrative Scenario 74 4.1.2 CP vs. UW Efficiency Analysis 76 4.1.3 Numerical Results 77 4.2 Sequences for UW and OOB Radiation 78 4.2.1 Orthogonal Polyphase Sequences 79 4.2.2 Waveform Engineering for UW Sequences combined with GFDM 79 4.2.3 Simulation Results for OOB Emission of UW-GFDM 81 4.3 Synchronization 82 4.3.1 Transmission over a Centralized MIMO Wireless Channel 82 4.3.2 Coarse Time Acquisition 83 4.3.3 CFO Estimation and Removal 85 4.3.4 Fine Time Acquisition 86 4.3.5 Simulation Results 88 4.4 Channel Estimation 92 4.4.1 MIMO UW-based LMMSE CE 92 4.4.2 Adaptive Filtering 93 4.4.3 Circular UW Transmission 94 4.4.4 Simulation Results 95 4.5 Equalization with Imperfect Channel Knowledge 96 4.5.1 UW-Free Equalization 97 4.5.2 Simulation Results 99 4.6 Summary 102 5 Conclusions and Perspectives 103 5.1 Main Outcomes in Short 103 5.2 Open Challenges 105 A Complementary Materials 107 A.1 Linear Algebra Identities 107 A.2 Proof of lower triangular Toeplitz channel matrix being defective 108 A.3 Calculation of noise-plus-interference covariance matrix for Pilot- and CPaided CE 108 A.4 Bock diagonalization of the effective channel for GFDM 109 A.5 Detailed complexity analysis of Sec. 3.4 109 A.6 CRLB derivations for the pdf (4.24) 113 A.7 Proof that (4.45) emulates a circular CIR at the receiver 114

Tor hidden services are anonymous servers of unknown location and ownership who can be accessed through any Tor-enabled web browser. They have gained popularity over the years, but still suer from major usability challenges due to their cryptographicallygenerated non-memorable addresses. In response to this difficulty, in this work we introduce the Onion Name System (OnioNS), a privacy-enhanced distributed DNS that allows users to reference a hidden service by a meaningful globally-unique veriable domain name chosen by the hidden service operator. We introduce a new distributed self-healing public ledger and construct OnioNS as an optional backwards-compatible plugin for Tor on top of existing hidden service infrastructure. We simplify our design and threat model by embedding OnioNS within the Tor network and provide mechanisms for authenticated denial-of-existence with minimal networking costs. Our reference implementation demonstrates that OnioNS successfully addresses the major usability issue that has been with Tor hidden services since their introduction in 2002.

La synthèse des protéines est un mécanisme central de la vie cellulaire dont la compré-hension est un enjeu pour la recherche biomédicale. Des phénomènes comme les erreurs programméesde la traduction eucaryote ou l’initiation par des structures IRES virales sont impliqués dans les processusde réplication de virus et de bactéries. Mieux appréhender ces processus est une étape essentiellepour aboutir au développement d’approches thérapeutiques innovantes. Les études en molécule uniquepermettent d’observer chaque système réactionnel individuellement et donnent accès à des évènementsasynchrones difficilement observables en mesure d’ensemble, tels la traduction de protéines.Ce manuscrit de thèse présente une approche d’étude de la traduction par un ribosome eucaryote(mammifère) en molécule unique. Nous observons les systèmes traductionnels grâce à des marqueursfluorescents liés à des oligonucléotides pouvant s’hybrider sur les séquences d’ARN messagers traduites.L’observation de ces marqueurs est faite par microscopie de fluorescence en réflexion totale (TIRFM),les ARNm étant accrochés à la surface de l’échantillon. En lisant l’ARNm, le ribosome détache lesmarqueurs, et leurs instants de départ nous permettent de remonter à la dynamique de traductionde ribosomes individuels. Cette méthode permet d’obtenir des données cinétiques statistiques surun grand nombre de systèmes traductionnels en parallèle pouvant alors être ajustées par des lois deprobabilité. Partant de ce principe, mes travaux de thèse ont eu pour objectif d’étendre nos expériencesà une nouvelle problématique biologique : l’étude des évènements non canoniques de la traductioneucaryote. Pour cela nous avons apporté les modifications et les optimisations nécessaires au dispositifet au protocole expérimental pour l’adapter à ces nouveaux enjeux.Nos mesures de la cinétique in vitro de l’élongation eucaryote ont mis à jour un délai dû à uneinitiation non-canonique. En effet, nous réalisons le recrutement du ribosome par l’ARNm grâce àune structure virale de type IRES. Dans nos conditions d’expérience, l’incorporation d’un acide aminéprend environ une seconde tandis que cette structure induit un retard à la traduction de plusieursdizaines de secondes. Nous avons réalisé une étude comparative de plusieurs de ces structures viraleset avons montré que le délai mesuré était une caractérisitique conservée dans le cadre de l’initiation noncanonique. Ce résultat ouvre des perspectives d’études cinétiques tant pour approfondir nos conclusionssur les IRES que pour aborder d’autres évènements non canoniques tel que le décalage de la phase delecture ou le franchissement du codon stop. / The synthesis of proteins is a central mechanism of cellular life whose understandingis an issue for biomedical research. Phenomena such as programmed errors of eukaryotic translation orinitiation by viral IRES structures are involved in virus and bacterial replication processes. A Betterunderstanding of these processes is an essential step towards the development of innovative therapeuticapproaches.Single molecule studies allow each reaction system to be observed individually and give accessto asynchronous events, such as protein translation, that are difficult to observe in overall measurements.This phD manuscript presents a single molecule approach to study translation by a eukaryotic(mammalian) ribosome.We observe the translational systems thanks to fluorescent primers linked to oligonucleotides thatare hybridized to the translated mRNA sequences. These markers are observed by Total InternalReflection Fuorescence Microscopy (TIRFM) ; with the mRNAs attached to the sample surface. Whilereading the mRNA, the ribosome detaches the primers, and their instants of departure give us access tothe translation dynamics of individual ribosomes. This method makes it possible to obtain statisticalkinetic data on a large number of parallel translational systems, which can then be fitted by probabilitylaws. On the basis of this principle, my phD work aimed at extending our experiments to a newbiological issue : the study of non-canonical events in eukaryotic translation. To this end, we havemade the modifications and optimizations necessary for the set-up and the experimental protocol toadapt them to these new challenges.Our measurements of the in vitro kinetics of eukaryotic elongation have revealed a delay due tonon-canonical initiation. Indeed, the ribosome are recruited on the mRNA thanks to a viral, IREStype structure. Under our experimental conditions, the incorporation of an amino acid takes aboutone second while this structure induces a translation delay of several tens of seconds. We carried outa comparative study of several of these viral structures and showed that the measured delay was acharacteristic preserved in the framework of the non-canonical initiation. This result opens up prospectsfor kinetic studies both to deepen our conclusions on IRES and to address other non-canonical eventssuch as programmed frameshifting or STOP codon readthrough.

Cette thèse a été consacrée à l'étude théorique et expérimentale du couplage contrôlé de la source de photon unique (SPS) aux structures photoniques multidimensionnelles, par l'utilisation de la méthode dite écriture directe par laser (DLW) par absorption ultra-faible à un photon (LOPA). La thèse est constitutée de trois parties principales suivantes:La première partie concerne le caractérisation et l'optimisation des dots quantiques (QD) colloïdaux CdSe/CdS. La dépendance de la longeur d'onde d'excitation a été étudiée. En utilisant une excitation à 532 nm, seul le core est excité et l’effet d'Auger est donc supprimé. Cette approche permet donc d'obtenir avec la suppression de la commutation intermittente et une source de photon unique très stable à température ambiante. Afin d’obtenir une meilleure performance des QDs sur une longue période, nous avons ensuite étudié l'influence du milieu dans lequel les QDs sont logés sur leurs propriétés optiques. En intégrant les QDs dans les matrices de polymère SU-8, nous avons montré que l'environnement polymérique permet non seulement de conserver de bonnes caractéristiques des QDs CdSe/CdS avec une photostabilité élevée, mais également de nous offrir une excellente accessibilité à la fabrication des structures en polymère contenant une particule unique.Dans la deuxième partie, la technique LOPA DLW est utilisée pour le couplage des QDs uniques dans diverses structures photoniques. Deux dispositifs, l’antenne du type ''pilier diélectrique'' sous-lambda et la membrane de réseau circulaire (cavité du type ''bulleye''), ont été étudiés théoriquement et expérimentalement pour améliorer l’émission du QD couplé en termes de l'émission spatiale et de l’émission radiative spontanée de l’émetteur.Dans la troisième partie, la manipulation de la SPS est démontrée en couplant le QD unique à des structures magnéto-photoniques multidimensionnelles. À l'aide d'un champ magnétique externe, le mouvement contrôlable d'un seul QD a été démontré dans un environnement fluidique. En contrôlant l'amplitude et l'orientation du champ magnétique externe, la position et l'orientation de la SPS à base d'un QD ont été manipulées à la demande. Les propriétés optiques, magnétiques et mécaniques des dispositifs magnéto-photoniques hybrides ont été étudiées pour montrer les capacités multifonctionnelles de telles structures. / The thesis has been devoted to study the controlled coupling of a colloidal quantum dot (QD) based single photon source (SPS) into multidimensional polymeric photonic structures by using low-one photon absorption (LOPA) direct laser writing (DLW) technique. The thesis consists of three main parts:The first part addresses the characteristic optimization of the CdSe/CdS based SPS. The excitation wavelength dependence of the QDs was investigated. By using 532 nm, only the core of the QD is excited with the suppression of the Auger effects. Thus, this approach allows for obtaining the suppression of fluorescence intermittency and a stable single-photon emission at ambient conditions. In order to obtain the long-term high fluorescence quality of the QDs, we then studied the influence of the local dielectric medium on the optical properties of the QDs. By incorporating the QDs into a photoresist (SU-8), we demonstrated that the polymeric environment not only enables the long-term preservation of the QD with high photostability but also provides us excellent accessibility to fabricate polymeric structures containing SPS.In the second part, the LOPA-based DLW is employed for the coupling of single QD into various photonic structures. Two devices including submicropillar dielectric antenna and 3D membrane bulleye cavity are theoretically and experimentally investigated to enhance the fluorescence emission of the single QD in terms of far-field angular radiation pattern and the spontaneous radiative emission of the emitter.In the third part, the manipulation of SPS is demonstrated by coupling the single QD into multidimensional magneto-photonic structures. With the aid of an external magnetic field, the controllable movement of the coupled QD was performed in the fluidic environment. The position and orientation of the SPS coupled in the structure were manipulated on demand. The mechanical, magnetic and optical properties of the device are investigated showing the multifunctional capabilities of magneto-photonic structures.

Dans la majeure partie des cas, la séparation en taille de molécules d'ADN, étape primordiale lors d'unséquençage, est réalisée par électrophorèse sur gels, inadaptée à la séparation de longues molécules : larecherche de techniques de séparation alternatives est donc primordiale. Nous avons utilisé unetechnologie de fabrication alternative, la Lithographie par Décalage de Phase, pour fabriquer des matricesd’obstacles de 80 à 500 nm de diamètre, de formes cylindrique ou ellipsoïdale. Ces matrices nous ontpermis de mener une étude des dynamiques de collision ADN-obstacle à l’échelle de la moléculeindividuelle, par la caractérisation des effets de l’actionnement (électrophorétique ou hydrodynamique), dela dimension et de la forme des obstacles sur ces dynamiques, impliquées dans le processus de séparationen taille. Nous montrons enfin la première séparation hydrodynamique de fragments d’ADN dans desréseaux d’obstacles nanométriques / In most cases, separation by size of DNA molecules, a crucial step for sequencing, is realized by gelelectrophoresis, unadapted to long molecule separation: it is consequently relevant to investigate alternativeseparation techniques. We have used an alternative fabrication technology, Phase Shift Lithography, tofabricate obstacle matrices which sizes range from 80 to 500 nm, with cylindrical or ellipsoidal shapes.These matrices allowed us to investigate DNA-obstacle collision dynamics at the single molecule scale, bythe caracterisation of actuation effects (electrophoretic or hydrodynamic) and of the size and shape of theobstacles on these dynamics, involved in the separation by size process. We finaly showed the firsthydrodynamic separation of DNA fragments into nanopilar matrices

The introduction of calculation methods for motion induced interruptions (MII) in 1984 introduced a new way to quantify human factors in addition to the motion sickness index (MSI). The 1990 Graham method for calculating MII uses a combination of a vessel's acceleration and roll to determine a "tipping" factor to calculate MII per minute. The Applebee-Baitis (AB) method considered that the motions are implicitly considered in accelerations, and therefore did not require roll to calculate MII. This thesis examines and analyzes the differences between the AB and Graham methods and compares their results for a unique hull form shape, a cylindrical capsule, in rough seas to determine which method is preferred. Two tests were performed by the Naval Surface Warfare Center, Carderock Division (NSWCCD) for the National Aeronautics and Space Administration (NASA) on the Orion Crew Exploration Vehicle (CEV) in post-splashdown conditions. A one-quarter scale model of the CEV was tested at the Aberdeen Test Center in Aberdeen, Maryland. Direct comparison of the analyzed data, MII sensitivity to location, and scaling analyses are examined and future work to further the application of MII calculation methods are proposed. The symmetry of the capsule leads to the assumption that roll and pitch-dominant MII calculations should be on the same order of magnitude. They are not because both MII methods only take roll-motions into account. The inclusion of both pitch and roll motions for the MII calculations is proposed as future work. The Graham method was found to be the more appropriate calculation because it is more conservative, and therefore preferred in the context of crew safety. / Master of Science

No description available.

A cut is a partition of a graph's nodes into two sets, and we say that an edge crosses the cut if it connects two nodes belonging to different sets. A maximum cut is a cut that maximises the number of crossing edges. We show that for any sufficiently small ε > 0 it is NP-hard to distinguish between graphs for which at least a fraction 1 - ε of all edges crosses the maximum cut and graphs for which at most a fraction 1 - 1.4568 ε of all edges crosses the maximum cut. The previous state of the art had a constant smaller than 1.375 in place of 1.4568. / Ett snitt är en partition av en grafs noder i två mängder, och vi säger att en kant korsar snittet om dess ändpunkter tillhör olika mängder. Ett maximalt snitt är ett snitt som maximerar antalet kanter som korsar snittet. Vi bevisar att det för alla tillräckligt små konstanter ε > 0 är NP-svårt att skilja mellan grafer för vilka minst en andel 1 - ε av alla kanter korsar det maximala snittet och grafer för vilka högst en andel 1 - 1.4568 ε av alla kanter korsar det maximala snittet. Detta är en förbättring jämfört med ett tidigare resultat som hade en konstant mindre än 1.375 istället för 1.4568.

Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

Thèse numérisée par la Direction des bibliothèques de l'Université de Montréal.