The recent decade has witnessed great advances in microarray and genotyping technologies which allow genome-wide single nucleotide polymorphism (SNP) data to be captured on a single chip. As a consequence, genome-wide association studies require the development of algorithms capable of manipulating ultra-large-scale SNP datasets. Towards this goal, this thesis proposes two SNP selection methods; the first using Independent Component Analysis (ICA) and the second based on a modified version of Fast Orthogonal Search. The first proposed technique, based on ICA, is a filtering technique; it reduces the number of SNPs in a dataset, without the need for any class labels. The second proposed technique, orthogonal search based SNP selection, is a multivariate regression approach; it selects the most informative features in SNP data to accurately model the entire dataset. The proposed methods are evaluated by applying them to publicly available gene SNP datasets, and comparing the accuracies of each method in reconstructing the datasets. In addition, the selection results are compared with those of another SNP selection method based on Principal Component Analysis (PCA), which was also applied to the same datasets. The results demonstrate the ability of orthogonal search to capture a higher amount of information than ICA SNP selection approach, all while using a smaller number of SNPs. Furthermore, SNP reconstruction accuracies using the proposed ICA methodology demonstrated the ability to summarize a greater or equivalent amount of information in comparison with the amount of information captured by the PCA-based technique reported in the literature. The execution time of the second developed methodology, mFOS, has paved the way for its application to large-scale genome wide datasets. / Thesis (Master, Computing) -- Queen's University, 2010-12-15 18:03:00.208

Due to their complementary characteristics, Global Positioning System (GPS) is usually integrated with standalone navigation devices like odometers and inertial measurement units (IMU). Recently, intensive research has focused on utilizing Micro-Electro-Mechanical-System (MEMS) grade inertial sensors in the integration because of their low cost. In this study, a reduced inertial sensor system (RISS) is considered. It comprises a MEMS grade single axis gyroscope, the vehicle built-in odometer, and two optional MEMS grade accelerometers. Estimation technique is needed to allow the data fusion of RISS and GPS. With adequate accuracy, Kalman filter (KF) fulfills this requirement if high-end inertial sensors are used. However, due to the inherent error characteristics of MEMS devices, MEMS-based RISS suffers from the non-stationary stochastic sensor errors and nonlinear inertial errors, which cannot be suppressed by KF alone. To solve the problem, Fast Orthogonal Search (FOS), a nonlinear system identification algorithm, is suggested in this research for modeling higher order RISS errors. FOS algorithm has the ability to figure out the system nonlinearity with a tolerance of arbitrary stochastic system noise. Its modeling results can then be used to predict the system dynamics. Motivated by the above merits, a KF/FOS module is proposed. By handling both linear and nonlinear RISS errors, this module targets substantial enhancement of positioning accuracy. To examine the effectiveness of the proposed technique, KF/FOS module is applied on RISS with GPS in a land vehicle for several road test trajectories. Its performance is compared to KF-only method, both assessed with respect to a high-end reference. To evaluate navigation algorithm in real-time vehicle application, a multi-sensor data logger is designed in this research to collect online RISS/GPS data. KF/FOS module is transplanted on an embedded digital signal processor as well. Both the off-line and online results confirm that KF/FOS module outperforms KF-only approach in positioning accuracy. They also demonstrate reliable real-time performance. / Thesis (Ph.D, Electrical & Computer Engineering) -- Queen's University, 2012-01-22 01:26:11.477

The thesis presents an implementation of a multiple input single output orthogonal frequency division multiplex (MISO OFDM) communication system testbed. The project was developed in order to evaluate whether the channel estimation algorithms developed by Dr Oyerinde [1] could be implemented in a real time communication system that uses today’s technology. This implementation based validation would help determine the practicality of algorithms and methods that promise better performance for communication systems from a simulation point of view. The benefits of using multiple orthogonal carriers are discussed as well as how an OFDM system works. The benefits of using multiple antennas at the transmitter, as opposed to using just one, are also discussed. The Alamouti scheme which allows space diversity to be achieved without the cost of having a lower data rate is presented. Modules common to all communication systems, such as those dedicated to synchronization, channel estimation, symbol detection and channel coding, are discussed. The different methods of synchronization for OFDM communication systems are presented and compared. The channel estimation algorithm developed by Dr Oyerinde is presented and is adopted for an indoor channel. Most of the system blocks and parameters used in the testbed are the same as those used in [1] in order to easily compare the results obtained by simulation and those obtained by implementation. The system bandwidth required for the project was too high for the processor chosen for the testbed. A qualitative evaluation of the practicality of Dr Oyerinde’s channel estimation algorithms was performed instead. From this evaluation it was derived that Dr Oyerinde’s non-iterative decision directed channel estimation algorithm was more suitable for real time non-iterative decision directed channel estimation communication systems than for iterative versions. Apart from processing demands that couldn’t be met, the other aspects of the project were implemented successfully. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2012.

In this thesis, a method of distortion analysis of electromagnetic field sensors using orthogonal polynomial subspaces is presented. The effective height of the sensor is viewed as the impulse response of a linear system. The impulse response corresponds to a linear transformation which maps every electromagnetic incident field waveform to a received voltage waveform. Hermite and Laguerre orthogonal polynomials are used as the basis sets for the subspace of incident electromagnetic field waveforms. Using the selected basis set, a transformation matrix is calculated for the sensors. The transformation matrices are compared to a reference transformation matrix as a measure of distortion. The transformation matrices can describe the sensor behavior up to a certain frequency range. The limits on this frequency range are investigated for both Hermite-Gauss and Laguerre functions. The unique property of Laguerre functions is used to prove that the transformation matrix has a particular pattern. This method is applied on case studied sensors both in computer simulation and measurements.

Material properties of polycrystalline aggregates should manifest the influence of crystallographic texture as defined by the orientation distribution function (ODF). A representation theorem on material tensors of weakly-textured polycrystals was established by Man and Huang (2012), by which a given material tensor can be expressed as a linear combination of an orthonormal set of irreducible basis tensors, with the components given explicitly in terms of texture coefficients and a number of undetermined material parameters. Man and Huang's theorem is based on the classical assumption in texture analysis that ODFs are defined on the rotation group SO(3), which strictly speaking makes it applicable only to polycrystals with (single) crystal symmetry defined by a proper point group. In the present study we consider ODFs defined on the orthogonal group O(3) and extend the representation theorem of Man and Huang to cover pseudotensors and polycrystals with crystal symmetry defined by any improper point group. This extension is important because many materials, including common metals such as aluminum, copper, iron, have their group of crystal symmetry being an improper point group. We present the restrictions on texture coefficients imposed by crystal symmetry for all the 21 improper point groups and we illustrate the extended representation theorem by its application to elasticity.

The synthesis of mesoporous silica thin films using surfactant templating typically leads to an inaccessible pore orientation, making these films not suitable for membrane applications. Recent advances in thin film synthesis provide for the alignment of hexagonal pores in a direction orthogonal to the surface when templated on chemically neutral surfaces. In this work, orthogonal thin film silica membranes are synthesized on alumina supports using block copolymer poly(ethylene glycol)-block-poly(propylene glycol)-block-poly(ethylene glycol) (P123) as the template. The orthogonal pore structure is achieved by sandwiching membranes between two chemically neutral surfaces, resulting in 90 nm thick films. Solvent flux of ethanol through the membrane demonstrates pore accessibility and suggests a silica pore size of approximately 10 nm. The permeability of ions and fluorescently tagged solutes (ranging from 4,000 to 70,000 Da) is used to demonstrate the membrane’s size selectivity characteristics. A size cut off occurs at 69,000 Da for the model protein BSA. By functionalizing the silica surface with a long chained alkyl group using n-decyltriethoxysilane (D-TEOS), the transport properties of the membranes can be altered. Contact angle measurements and FTIR results show the surface to be very hydrophobic after functionalization. Solvent flux of ethanol through the silica thin film membrane is similar before and after functionalization, but water flux decreases. Thin film silica membranes show much promise for applications in catalysis, bio-sensing, and affinity separations.

Various multiuser schemes have been proposed to efficiently utilize the available bandwidth while ensuring an acceptable service delivery and flexibility. The multicarrier CDMA became an attractive solution to the major challenges confronting the wireless communication system. However, the scheme is plagued with multiple access interference (MAI), which causes conspicuous performance deterioration at the receiver. A low-complexity multiuser scheme called the Interleave Division Multiple Access (IDMA) was proposed recently as a capable solution to the drawback in the multicarrier CDMA scheme. A combined scheme of OFDM-IDMA was later introduced to enhance the performance of the earlier proposed IDMA scheme. The multicarrier IDMA scheme therefore combats inter-symbol interference (ISI) and MAI effectively over multipath with low complexity while ensuring a better cellular performance, high diversity order, and spectral efficiency. Major studies on the OFDM-IDMA scheme emphasis only on the implementation of the scheme in a perfect scenario, where there are no synchronization errors in the system. Like other multicarrier schemes, the OFDM-IDMA scheme however suffers from carrier frequency offset (CFO) errors, which is inherent in the OFDM technique. This research work therefore examines, and analyzes the effect of synchronization errors on the performance of the new OFDM-based hybrid scheme called the OFDM-IDMA. The design of the OFDM-IDMA system developed is such that the cyclic prefix duration of the OFDM component is longer than the maximum channel delay spread of the multipath channel model used. This effectively eliminates ISI as well as timing offsets in the system. Since much work has not been done hitherto to address the deteriorating effect of synchronization errors on the OFDM-IDMA system, this research work therefore focuses on the more challenging issue of carrier frequency synchronization at the uplink. A linear MMSE-based synchronization algorithm is proposed and implemented. The proposed algorithm is a non-data aided method that focuses on the mitigation of the ICI induced by the residual CFOs due to concurrent users in the multicarrier system. However, to obtain a better and improved system performance, the Kernel Least Mean Square (KLMS) algorithm and the normalized KLMS are proposed, implemented, and effectively adapted to combat the degrading influence of carrier frequency offset errors on the OFDM-IDMA scheme. The KLMS synchronization algorithm, which involves the execution of the conventional Least Mean Square (LMS) algorithm in the kernel space, utilizes the modulated input signal in the implementation of the kernel function, thereby enhancing the efficacy of the algorithm and the overall output of the multicarrier system. The algorithms are applied in a Rayleigh fading multipath channel with varying mobile speed to verify their effectiveness and to clearly demonstrate their influence on the performance of the system in a practical scenario. Also, the implemented algorithms are compared to ascertain which of these algorithms offers a better and more efficient system performance. Computer simulations of the bit error performance of the algorithms are presented to verify their respective influence on the overall output of the multicarrier system. Simulation results of the algorithms in both slow fading and fast fading multipath scenarios are documented as well. / Thesis (M.Sc.Eng.)-University of KwaZulu-Natal, Durban, 2013.

This thesis presents strategies for constructing multi-functional biomaterials based on hyaluronan (HA) derivatives for various biomedical applications, such as drug delivery, tissue regeneration, and imaging biomaterials. The aim of this study is to improve the functionalities of HA biomaterials as well as simplify the preparation procedures.  Native HA polymer contains D-glucuronic acid residue with a carboxyl group per disaccharide unit that can be easily modified by carbodiimide-mediated amidation reaction. Therefore, we have designed a series of orthogonal groups (hydrazide, carbazate, aldehyde, and thiol) that can be linked to HA under mild conditions using the carbodiimide chemistry. Multiple functionalities can be introduced to the obtained HA derivatives via chemoselective “click”-type transformations.   The modified HA derivatives were used for the preparation of either nanogel particles (NPs) or bulk hydrogels. Due to “click” character of the reactions used, structural HA transformations were performed with high fidelity on different scales including molecular (polymers), nanometer (NPs), and a visible scale (bulk hydrogels). By linking pyrene or camptothecin to hydrophilic HA backbone, amphiphilic polymers were obtained and utilized as drug delivery carriers or prodrugs, respectively. Subsequently, physically loaded drug (doxorubicin) could be released upon degradation of HA carriers, while the chemically linked camptothecin was released intact by a thiol-triggered cleavage reaction. Bisphosphonated HA (HA-BP) polymers were prepared to induce hydrogel scaffold bio-mineralization for bone regeneration application. Moreover, we could recruit strong binding capacity of bisphosphonate (BP) groups to calcium ions for the formation of physically crosslinked HA-BP gel upon simple mixing of the polymer and calcium phosphate nanoparticle components. This gel was more stable in vivo compared to hydrazone crosslinked HA gels. Furthermore, the hydrogel composed of fluorine-19 (19F) linked HA polymer was successfully observed by both 1H and 19F MR imaging.         In conclusion, the presented herein study describes new approaches for building up multi-functional biomaterials from the HA-based blocks. The utilization of carbodiimide and click chemistries along with the enzymatic degradation of HA allowed simple and efficient interconversion between HA macromolecules, nanoparticles and macroscopic hydrogels. These HA-based biomaterials show high potential for use in the fields of drug delivery, bone regeneration, and imaging techniques.

In MALDI, for laser fluences below the saturation point the ion yield per shot follows a cubic dependence on the irradiated area, leading to a conclusion that smaller spots produce overall less ions and therefore are less viable. However, Qiao et al. showed that by decreasing the laser spot size it is possible to raise the saturation point, and thus increase the ion yield per unit area, also known as sensitivity. Here we explore laser spots below 10 micrometer diameter to determine whether they offer any practical advantage. We show that sensitivity is greater for a flat-top 3-4 micrometer spot than for a 10 micrometer spot. The sensitivity is greater for a Gaussian-like 3-5 micrometer spot than for flat-top 5-25 micrometer spots. We also report for the first time sensitivity versus theoretical fluence profile for a Gaussian-like beam focused to a spot of 3-5 micrometer.

In this thesis, a method of distortion analysis of electromagnetic field sensors using orthogonal polynomial subspaces is presented. The effective height of the sensor is viewed as the impulse response of a linear system. The impulse response corresponds to a linear transformation which maps every electromagnetic incident field waveform to a received voltage waveform. Hermite and Laguerre orthogonal polynomials are used as the basis sets for the subspace of incident electromagnetic field waveforms. Using the selected basis set, a transformation matrix is calculated for the sensors. The transformation matrices are compared to a reference transformation matrix as a measure of distortion. The transformation matrices can describe the sensor behavior up to a certain frequency range. The limits on this frequency range are investigated for both Hermite-Gauss and Laguerre functions. The unique property of Laguerre functions is used to prove that the transformation matrix has a particular pattern. This method is applied on case studied sensors both in computer simulation and measurements.