In this thesis, we design, develop and implement parallel algorithms for irregular problems on heterogeneous multi-core architectures. Irregular problems exhibit random and unpredictable memory access patterns, poor spatial locality and input dependent control flow. Heterogeneous multi-core processors vary in: clock frequency, power dissipation, programming model (MIMD vs. SIMD), memory design and computing units, scalar versus vector units. The heterogeneity of the processors makes designing efficient parallel algorithms for irregular problems on heterogeneous multicore processors challenging. Techniques of mapping tasks or data on traditional parallel computers can not be used as is on heterogeneous multi-core processors due to the varying hardware. In an attempt to understand the efficiency of futuristic heterogeneous multi-core architectures on applications we study several computation and bandwidth oriented irregular problems on one heterogeneous multi-core architecture, the IBM Cell Broadband Engine (Cell BE). The Cell BE consists of a general processor and eight specialized processors and addresses vector/data-level parallelism and instruction-level parallelism simultaneously. Through these studies on the Cell BE, we provide some discussions and insight on the performance of the applications on heterogeneous multi-core architectures. Verifying these experimental results require some performance modeling. Due to the diversity of heterogeneous multi-core architectures, theoretical performance models used for homogeneous multi-core architectures do not provide accurate results. Therefore, in this thesis we propose an analytical performance prediction model that considers the multitude architectural features of heterogeneous multi-cores (such as DMA transfers, number of instructions and operations, the processor frequency and DMA bandwidth). We show that the execution time from our prediction model is comparable to the execution time of the experimental results for a complex medical imaging application.

This dissertation presents the contributions and the research conducted in developing and implementing Microwave Tomography (MWT) systems. MWT is an imaging modality which aims to interrogate an object of interest by microwave energy, and quantitatively “find” the interior spatial distribution of its dielectric properties using field measurements taken outside the object. Due to the inherent non-linearity of the MWT problem, a substantial amount of electromagnetic scattering data is required to ensure a robust inversion and quantitatively accurate imaging results. This research benefits a variety of applications including biomedical imaging, industrial non-destructive testing, and security applications. Developing a MWT system, requires many critical components including the bandwidth and polarization purity of the collected fields as well as calibration of the fields scattered by the object of interest. Two generations of MWT systems were designed, implemented, calibrated and tested at the University of Manitoba (UM). These systems aim different approaches for near-field measurements which are referred to as the direct and indirect methods. With regard to the antenna design, a novel methodology applicable to broadband planar antennas is introduced. This technique is based on a combination of field modelling, herein, the finite element method and transmission line modelling. In the first generation of the UM MWT systems, a suitable antenna system was utilized. The system under study was a prototype, where twenty-four co-resident antennas encircle the object of interest to directly measure the fields. In the second generation of the UM MWT systems, the feasibility of using a novel technique to indirectly measure the fields by a secondary array of near-field scatterer probes was studied. The technique is based on the Modulated Scatterer Technique (MST). In this system, antennas are called ``collectors", since the role of antennas are changed to collecting probes' scattered fields. A number of PIN diodes were utilized to activate the probes. Finally, the capability of the probe system was investigated and its performance with the previously constructed tomography systems was compared. Various dielectric phantoms were utilized to test the accuracy of the systems.

In this thesis, we design, develop and implement parallel algorithms for irregular problems on heterogeneous multi-core architectures. Irregular problems exhibit random and unpredictable memory access patterns, poor spatial locality and input dependent control flow. Heterogeneous multi-core processors vary in: clock frequency, power dissipation, programming model (MIMD vs. SIMD), memory design and computing units, scalar versus vector units. The heterogeneity of the processors makes designing efficient parallel algorithms for irregular problems on heterogeneous multicore processors challenging. Techniques of mapping tasks or data on traditional parallel computers can not be used as is on heterogeneous multi-core processors due to the varying hardware. In an attempt to understand the efficiency of futuristic heterogeneous multi-core architectures on applications we study several computation and bandwidth oriented irregular problems on one heterogeneous multi-core architecture, the IBM Cell Broadband Engine (Cell BE). The Cell BE consists of a general processor and eight specialized processors and addresses vector/data-level parallelism and instruction-level parallelism simultaneously. Through these studies on the Cell BE, we provide some discussions and insight on the performance of the applications on heterogeneous multi-core architectures. Verifying these experimental results require some performance modeling. Due to the diversity of heterogeneous multi-core architectures, theoretical performance models used for homogeneous multi-core architectures do not provide accurate results. Therefore, in this thesis we propose an analytical performance prediction model that considers the multitude architectural features of heterogeneous multi-cores (such as DMA transfers, number of instructions and operations, the processor frequency and DMA bandwidth). We show that the execution time from our prediction model is comparable to the execution time of the experimental results for a complex medical imaging application.

Cette thèse porte sur la caractérisation non destructive de structures 2D magnéto-diélectriques inhomogènes complexes. L’ensemble des étapes allant de l’expérience au traitement du problème inverse est traité. Dans un premier temps, un modèle direct reliant le champ diffusé aux propriétés électromagnétiques du matériau a été mis en place. Ce modèle requiert des calculs par éléments finis de la propagation de l’onde électromagnétique, en présence de l’objet observé lorsque celui-ci est positionné sur un support métallique. Une validation expérimentale a été réalisée via la mise en place d'un banc de mesure multi statique. Différentes étapes d'ajustements et d'étalonnages ont permis la réduction du bruit de mesure ainsi que des biais. L’inversion est traitée principalement par une approche linéaire, avec un choix attentif de la valeur des hyper paramètres qui y sont associés. Une fois les outils mis en place, six études ont été réalisées pour la validation de notre système d’imagerie 2D des propriétés électromagnétiques de matériaux magnéto-diélectriques inhomogènes. Cela comprend l’évaluation des incertitudes de mesure, de la résolution spatiale, la mesure de différents matériaux magnétiques et l’utilisation de différents supports à géométries variées. L’ensemble des résultats expérimentaux réalisés se place dans une hypothèse de géométrie 2D. C’est pourquoi, nous avons ensuite orienté nos travaux vers la recherche d’un design innovant permettant de faire évoluer le banc de mesure en un dispositif d’imagerie 3D. Dans cette perspective, une source secondaire vient se déplacer proche de la cible pour acquérir de l’information selon la troisième dimension. / The subject of this thesis is the non-destructive characterization of complex inhomogeneous magneto-dielectric structures. Successively, the experimental developments, the modelling and the data treatments stages are addressed. A forward model that links the scattered field to the electromagnetic properties is established. This model requires some finite element computations in order to estimate the propagation of the electromagnetic wave in presence of the magneto-dielectric object which is glued on a metallic support. A multistatic bench has been designed and constructed in order to collect measured scattered fields. Several adjustments and calibration procedures have been carried out to reduce the measurement noise and biases. Next, the inverse problem has been dealt with, in order to retrieve the electromagnetic properties of the samples, from the measured scattered field. The inverse problem is mainly solved with a linear approach, with a careful selection of the hyperparameters. Once the system has been fine tuned, six studies have been realized to validate our 2D imaging system. The assessment of the measurement uncertainty, the evaluation of the spatial resolution, the characterization of various magnetics materials and the use of different supports with variable geometries have been performed. So far, all the developments were done under a 2D hypothesis. That is why, we have then focused our research on the design of a 3D innovative imaging setup. To this end, a secondary source moving close to the target has been added in order to gain information in the third direction. A numerical study has been performed to assess the expected performances of this new setup.

This thesis presents the research results on the development of a microwave tomography imaging algorithm capable of reconstructing the dielectric properties of the unknown object. Our focus was on the theoretical aspects of the non-linear tomographic image reconstruction problem with particular emphasis on developing efficient numerical and non-linear optimization for solving the inverse scattering problem. A detailed description of a novel microwave tomography method based on frequency dependent finite difference time domain, a numerical method for solving Maxwell's equations and Genetic Algorithm (GA) as a global optimization technique is given. The proposed technique has the ability to deal with the heterogeneous and dispersive object with complex distribution of dielectric properties and to provide a quantitative image of permittivity and conductivity profile of the object. It is shown that the proposed technique is capable of using the multi-frequency, multi-view, and multi-incident planer techniques which provide useful information for the reconstruction of the dielectric properties profile and improve image quality. In addition, we show that when a-priori information about the object under test is known, it can be easily integrated with the inversion process. This provides realistic regularization of the solution and removes or reduces the possibility of non-true solutions. We further introduced application of the GA such as binary-coded GA, real-coded GA, hybrid binary and real coded GA, and neural-network/GA for solving the inverse scattering problem which improved the quality of the images as well as the conversion rate. The implications and possible advantages of each type of optimization are discussed, and synthetic inversion results are presented. The results showed that the proposed algorithm was capable of providing the quantitative images, although more research is still required to improve the image quality. In the proposed technique the computation time for solution convergence varies from a few hours to several days. Therefore, the parallel implementation of the algorithm was carried out to reduce the runtime. The proposed technique was evaluated for application in microwave breast cancer imaging as well as measurement data from university of Manitoba and Institut Frsenel's microwave tomography systems.

This thesis presents the research results on the development of a microwave tomography imaging algorithm capable of reconstructing the dielectric properties of the unknown object. Our focus was on the theoretical aspects of the non-linear tomographic image reconstruction problem with particular emphasis on developing efficient numerical and non-linear optimization for solving the inverse scattering problem. A detailed description of a novel microwave tomography method based on frequency dependent finite difference time domain, a numerical method for solving Maxwell's equations and Genetic Algorithm (GA) as a global optimization technique is given. The proposed technique has the ability to deal with the heterogeneous and dispersive object with complex distribution of dielectric properties and to provide a quantitative image of permittivity and conductivity profile of the object. It is shown that the proposed technique is capable of using the multi-frequency, multi-view, and multi-incident planer techniques which provide useful information for the reconstruction of the dielectric properties profile and improve image quality. In addition, we show that when a-priori information about the object under test is known, it can be easily integrated with the inversion process. This provides realistic regularization of the solution and removes or reduces the possibility of non-true solutions. We further introduced application of the GA such as binary-coded GA, real-coded GA, hybrid binary and real coded GA, and neural-network/GA for solving the inverse scattering problem which improved the quality of the images as well as the conversion rate. The implications and possible advantages of each type of optimization are discussed, and synthetic inversion results are presented. The results showed that the proposed algorithm was capable of providing the quantitative images, although more research is still required to improve the image quality. In the proposed technique the computation time for solution convergence varies from a few hours to several days. Therefore, the parallel implementation of the algorithm was carried out to reduce the runtime. The proposed technique was evaluated for application in microwave breast cancer imaging as well as measurement data from university of Manitoba and Institut Frsenel's microwave tomography systems.