This thesis deals with the problem of practical application of FDTD technique for simulation of image formation in coherence controlled holographic microscope. Various ways for obtaining scattering matrices are explored in detail and the optimal technique based on a rigorous calculation of the far field is proposed. The scattering matrix, containing information about the observed sample, is subsequently used in analytic calculation of holographic signal; two levels of approximation of pupil function are also evaluated. The results are compared with a traditional approach based on Rytov approximation resulting in specification of the parameter domain where the approximation is applicable. Based on the simulations of the microscope, the dependence of axial resolution on apertures of the objective and the condenser and sensitivity of the signal to changes of refractive index of the sample is also studied.

The wireless industry is such that suppliers of antennas have to adapt their designs to requirement changes over a period of just a few months. In these short design cycles time is crucial. Radiation pattern testing of the antennas at various points in this design cycle are nowadays mostly done using spherical near-field techniques, where the tangential electric field is acquired over an imaginary sphere close to, and surrounding, the antenna under test, and this data then transformed into a far-zone radiation pattern. There are some applications where acquisition over a rotationally symmetric surface other than a spherical one would not only reduce test times, but allow equipment cost reductions as well. However, near-field to far-field transformations for finite non-spherical measurement surface shapes are not available. Such a transformation is proposed, implemented and validated in this thesis. It uses the method of moments, customized to a rotationally symmetric surface (body of revolution) to effect this transformation.

No description available.

In the last two decades, two major earthquakes have occurred in Japan: the 1995 Kobe earthquake and the 2011 Great East Japan earthquake. In the 2011 Great East Japan earthquake, many bridge structures were destroyed by the tsunamis, but it is interesting to study the ground motion induced damage and also how this earthquake differed from the one in 1995. In this thesis, the seismic response of a bridge designed according to the current Japanese Design Specifications was evaluated when it is subjected to near-field ground motions recorded during the 1995 Kobe earthquake and far-field ground motions recorded during the 2011 Great East Japan earthquake. For this purpose, a series of nonlinear dynamic response analysis was conducted and the seismic performance of the bridge was verified in terms of its displacement and ductility demand. It was found from the dynamic response analysis that the seismic response of the target bridge when subjected to the ground motions from the 2011 Great East Japan earthquake was smaller than during the 1995 Kobe earthquake. Although the ground motions from the 2011 Great East Japan earthquake were very strong, they were not as strong as the ground motions from the 1995 Kobe earthquake. The results obtained in this thesis clarify the validity of the Type I and Type II design ground motions. The target bridge used in this thesis was designed according to the post-1990 design specifications and showed limited nonlinear response when subjected to the different ground motions which shows how efficient the enhancement of the seismic performance of bridges has been since the 1990’s.

Advanced Encryption Standard (AES) is a widely accepted encryption algorithm used in Internet-of-Things (IoT) devices such as Bluetooth devices. Although the implementation of AES is complicated enough, attackers can still acquire the cryptographic information generated from the AES execution to perform Side-Channel Attack (SCA). There are two commonly used types of SCA, which are power based attack and Electromagnetic (EM) based attack. However, the acquisition of both power traces and EM near-field traces require close physical contact to the victim devices, which is difficult to attack a well-protected system. In this thesis, we exploit the far-field EM propagation property and train several Deep Learning (DL) models to attack tinyAES algorithm implemented on the victim Bluetooth chip nRF52832 mounted on Nordic nRF52 DK at the distance up to 50cm. To simulate the real attacking scenario, we train our DL models on one nRF52 DK at 30cm and attack another same board at the distance 5cm, 15cm, 30cm and 50cm respectively in an office environment. We restrict the number of attacking traces to 7000. The key byte of all of cases can be recovered successfully by Convolution Neuron Network (CNN) and the best test only need 1848 traces. Our contributions are: (1).We prove it is feasible to attack Bluetooth chip running AES at variation distance by DL; (2).We compare our DL model performance with the classical correlation analysis and find correlation analysis takes far more traces than DL; (3).We propose several countermeasures to protect against the far-field EM SCA. / Advanced Encryption Standard (AES) är en allmänt accepterad krypteringsalgoritm som används i Internet-of-Things (IoT) -enheter som Bluetooth-enheter. Även om implementeringen av AES är tillräckligt komplicerad kan angriparna fortfarande förvärva den kryptografiska informationen som genererats från AES-utförandet för att utföra Side-Channel Attack (SCA). Det finns två vanligt förekommande typer av SCA, som är kraftbaserad attack och elektro-magnetisk (EM) baserad attack. Emellertid kräver förvärv av både strömspår och EM-fältspår nära fysisk kontakt med offeranordningarna, vilket är omöjligt att attackera ett välskyddat system. I den här avhandlingen utnyttjar vi EM-förökningsegenskapen för fjärrfältet och utbildar flera Deep Learning (DL) -modeller för att attackera litenAES- algoritm implementerad på offret Bluetooth-chip nRF52832 monterat på Nordic nRF52 DK på avståndet upp till 50 cm. För att simulera det verkliga angreppsscenariot utbildar vi våra DL-modeller på en nRF52 DK vid 30 cm och attackerar en annan samma skiva på avståndet 5 cm, 15 cm, 30 cm respektive 50 cm i en kontorsmiljö. Vi begränsar antalet attackerande spår till 7000. Nyckelbyte i alla fall kan framgångsrikt återvinnas av Convolution Neuron Network (CNN) och det bästa testet behöver endast 1848 spår. Våra bidrag är: (1). Vi bevisar att det är möjligt att attackera Bluetooth-chip som kör AES på variation avstånd av DL; (2). Vi jämför våra DL-modellprestanda med den klassiska korrelationsanalysen och finner korrelationsanalys tar mycket fler spår än DL;(3). Vi tillhandahåller flera motåtgärder mot EM-SCA.

No description available.

This thesis focuses on the application of electromagnetic inverse source techniques to characterize antennas using phaseless (amplitude-only) near-field (NF) measurement data. Removing the need to measure phase reduces the overall cost of the measurement apparatus since simple power meters can be used instead of expensive vector network analyzers. It has also been shown in the literature that a phaseless approach can improve the accuracy of the calculated far-field (FF) pattern in the presence of probe positioning errors compared to the amplitude-and-phase approach. A brief discussion on the state-of-the-art methods for characterizing antennas using phaseless near-field measurement data is presented. Two general approaches used most often to perform near-field to far-field (NF-FF) transformations, namely modal expansion and source reconstruction, are explained in detail for scenarios with and without phase information. A phaseless source reconstruction method (SRM) is the primary focus of this work. The SRM is an application of an electromagnetic inverse source technique and therefore, the complexities of solving the associated ill-posed inverse source problem are discussed. The application of the SRM to spherical and planar measurement geometries are presented along with the concerns regarding regularization resulting from discretizing the ill-posed system. A multiplicative regularization (MR) scheme originally developed for inverse scattering is adapted to suit the nonlinear cost functional for the phaseless planar measurement case and the mathematical framework is derived in detail. The resulting MR-SRM is fully automated and incorporates adaptive regularization. The developed algorithms are evaluated using several examples with synthetic phaseless NF data demonstrating the benefits and limitations of the source reconstruction method and the multiplicative regularization scheme. The application of the SRM to antenna diagnostics using phaseless NF data is also shown. Finally, the developed planar algorithms are tested with experimentally collected phaseless measurement data to demonstrate their potential as suitable antenna characterization techniques that can be of interest to the antenna measurement community. / October 2016

Les nanocavités à cristal photonique ont été largement étudiées au cours de la dernière décennie du fait de leur aptitude à fortement confiner la lumière (faible volume modal) et de leurs faibles pertes optiques (grand facteur de qualité). Parmi le grand nombre de géométries proposées, nous nous intéressons ici au cas de la cavité L3 étudiée par Noda et al. (trois trous manquants dans la direction K du réseau triangulaire sous-jacent) utilisée dans plusieurs applications et notamment pour la réalisation de nano-lasers et d’interrupteurs optiques. Cependant, l’injection ou l’extraction de lumière dans de telles nanocavités s’avère extrêmement difficile du fait de la diffraction importante dont souffrent ces structures. Différentes approches en champ proche ont été récemment développées et notamment le couplage évanescent utilisant des guides d’onde nanostructurés ou des fibres optiques étirées. Dans le but de pallier à la faible efficacité de couplage à l’espace libre, nous développons une conception récemment proposée par De Rossi et al. afin de changer radicalement le profil du diagramme de rayonnement. Cette approche utilise la méthode de repliement des bandes qui consiste à introduire au sein d’un réseau triangulaire de période (a), un sous-réseau de trous de période (2a) qui améliore considérablement l’efficacité de couplage dans la direction verticale. Bien que certaines mesures de l’efficacité de collection et du coefficient de qualité aient déjà été mentionnées dans la littérature, aucune mesure directe des diagrammes de rayonnement de ces nanocavités n’a été réalisée jusqu’alors. Nous étudions dans ce travail différents types de nanocavités et de molécules L3 à cristaux photoniques présentant des profils de champ lointain optimisés. Les diagrammes de rayonnement de cavités non-repliées et repliées incorporées dans des membranes actives suspendues en InP sont systématiquement mesurés et comparés. Un bon accord entre les simulations numériques et les diagrammes de champ lointain mesurés expérimentalement est obtenu, montrant des lobes d’émission très directionnels le long de la normale à l’échantillon. En outre, des expériences de couplage à l’espace libre ont été réalisées montrant des efficacités de couplage d’environ 15% pour des coefficients de qualité supérieurs à 10 000. Ces résultats valident ainsi la technique de repliement des bandes dans les cavités L3 qui, une fois repliées, conservent un faible volume modal et un coefficient de qualité élevé ainsi qu’une grande efficacité de couplage à l’espace libre, à la fois dans les configurations nanocavité unique et nanocavités couplés. Nous montrons aussi expérimentalement que l’écart spectral inter-modal dans deux cavités L3 couplées de manière évanescente peut être contrôlé grâce à l’ingénierie de la barrière photonique. La « barrière de potentiel » est formée par les trous d’air séparant les deux cavités. L’écart en fréquence entre les modes peut être fortement réduit et augmentée via une diminution ou une augmentation du rayon des trous de la rangée centrale de la barrière jusqu’à ∼ −30% ou ∼ 30% de sa valeur initiale. En outre, le signe de la l’écart spectral entre les modes peut être inversé de telle sorte que le mode fondamental peut être soit symétrique ou anti-symétrique et ce, sans modifier ni la géométrie de la cavité, ni la distance inter-cavité. / Photonic crystal (PhC) nanocavities have been intensively investigated during the last decade due to their capabilities of achieving tight light confinement and low optical losses simultaneously. Among the different geometries, the cavity proposed by Noda et al., namely a L3 cavity (three holes missing in the K direction of the underlying triangular lattice) with shifted end-holes has been widely used in several applications including laser emission and switching devices. However, input/output free space light coupling of such nanocavities is quite challenging. In this regard, near field coupling schemes have been recently developed, such as evanescent coupling using tapered optical fibers. In order to overcome the poor free space coupling, a new cavity design has been recently proposed by De Rossi et al. that totally changes the radiation pattern. This is based on a band folding approach introducing a modulation of the holes size at twice the period of the underlying PhC, which considerably increases the coupling efficiency in the vertical direction. While some measurements of the Q-factor and coupling efficiency were performed, no direct characterization of the far-field of such cavities has been performed so far. In this work we have studied different types of L3 photonic crystal cavities and L3 photonic molecules with optimized far-field profiles. Radiation patterns from « folded » and « unfolded » cavities incorporated in suspended InP active membranes were systematically measured and compared. Good agreement between simulations and experimental far-field patterns has been found, demonstrating highly directional emission lobes along the sample normal. Furthermore, free space input coupling experiments have been performed showing coupling efficiency of about 15% of contrast with quality factors exceeding 10 000. These results validate the « folded » L3 cavities as good candidates for small volume and high Q cavities with efficient free space coupling, either in single or coupled cavity configurations. We also experimentally show that the mode splitting in two-evanescently coupled Photonic Crystal L3 cavities can be controlled through photonic barrier engineering. The « potential barrier » is formed by the air-holes in between the two cavities. By changing the hole radius of the central row in the barrier up to ∼ 30% or down to ∼ −30% , the frequency splitting can be strongly increased or reduced. Moreover, the sign of the splitting can be reversed in such a way that the fundamental mode can be either the symmetric or the anti-symmetric one without altering neither the cavity geometry nor the inter-cavity distance.

Gallium nitride (GaN)-based light-emitting diodes (LEDs) with cylindrical nanostructures have been the subject of significant research interest in the past decade, due to the potential of such structures to increase light extraction efficiency and deliver highly directional light emission. Nanorod LEDs, where the light emission is within the nanocylinder, have the additional potential to increase internal quantum efficiency and emit in colours previously thought impractical with GaN-based LEDs. Optical modelling of nanostructured LEDs is usually carried out using finite-difference time-domain methods, which are computationally intensive and do not always provide sufficient insight into the physics underlying the simulation results. This thesis proposes an intuitive, quasi-analytic method based on modal expansion. It is found that it is possible to calculate the far field diffraction patterns of all guided modes supported by a single nanorod, with full consideration of Fabry-Perot effects, in minutes using a standard office desktop computer. Focus is placed on the case of a nanorod of radius 140 nm, for which angular photoluminescence measurements were available to provide a means of validating the model. Consideration of the guided modes alone provides a compelling explanation for gross features in the measured data where none previously existed. It is shown that, using a standard equation from a textbook, it is possible to calculate how much each of the guided and radiation modes of a single nanorod is excited by a Hertzian dipole of known position and orientation with respect to the nanorod geometry. When interference between these modes is considered, it is possible to calculate the total far field angular emission pattern due to that dipole. Comparing these patterns with photoluminescence measurements allows one to infer the locations and orientations of dipole current sources; the results are found to be consistent with those of cathodoluminescence studies.

This thesis presents detailed fuel spray investigations relating to an automotive Gasoline Direct Injection (GDI) pressure-swirl injector, employing a combination of numerical and physical analyses. The emphasis is placed on the near-nozzle in recognition that all later flow processes are dominated by this critical region. To enable the technology to maximise its potential, it is essential to further our understanding of the fundamental flow physics that govern the injection process, which remain largely unknown. The complexity of the spray process has led to many avenues of research. Simplified models are particularly suitable for parametric studies, allowing fast computation of some of the most important design parameters, such as nozzle discharge coefficient, cone angle and initial velocity. More complex methods such as Computational Fluid Dynamics (CFD) offer significantly more detail including the temporal and spatial evaluation of the flow field and fuel distribution, but at the cost of often lengthy computational time, and the need to tune models against physical evidence. Unfortunately none are able to describe all aspects of the injection event simultaneously. A considerable body of existing experimental data gathered under atmospheric conditions has been condensed and carefully presented to provide a comprehensive picture of injector operation. This comprises global spray performance data, spray imaging, and droplet velocity and size maps as a function of time after the Start Of Injection (SOl). These serve to provide a means to develop physical models and to correlate model predictions. Particular attention is drawn to the challenges faced by numerical methods to successfully predict the complex spray behaviour. A fundamental computational study employing the Volume Of Fluid (VOF) method describes droplet break-up under controlled conditions. By varying the Weber number of the flow the expected break-up mechanisms are recovered, and the numerics and case set-up tuned to offer a practical balance between the resource burden and solution accuracy. This paved the way to a detailed 3-D transient analysis of the near-nozzle region of a pressure-swirl injector. Computed results clearly identify the consecutive phases of the fuel spray development, from the initial unsteady jet through to the stable, swirling hollow cone formation. Comparison with experimental measurements revealed that the computational approach is able to capture the main qualitative features of the spray process.