Please read the abstract in the front pages of the file named 00dissertation / Dissertation (MEng (Mechanical))--University of Pretoria, 2007. / Mechanical and Aeronautical Engineering / unrestricted

Intersymbol interference (ISI) and co-channel interference (CCI) are two major obstacles to high speed data transmission in wireless cellular communications systems. Unlike thermal noise, their effects cannot be removed by increasing the signal power and are time-varying due to the relative motion between the transmitters and receivers. Space-time processing offers a signal processing framework to optimally integrate the spatial and temporal properties of the signal for maximal signal reception and at the same time, mitigate the ISI and CCI impairments. In this thesis, we focus on the development of this emerging technology to combat the undesirable effects of ISI and CCL We first develop a convenient mathematical model to parameterize the space-time multipath channel based on signal path power, directions and times of arrival. Starting from the continuous time-domain, we derive compact expressions of the vector space-time channel model that lead to the notion of block space-time manifold, Under certain identifiability conditions, the noiseless vector-channel outputs will lie on a subspace constructed from a set. of basis belonging to the block space-time manifold. This is an important observation as many high resolution array processing algorithms Can be applied directly to estimate the multi path channel parameters. Next we focus on the development of semi-blind channel identification and equalization algorithms for fast time-varying multi path channels. Specifically. we develop space-time processing algorithms for wireless TDMA networks that use short burst data formats with extremely short training data. sequences. Due to the latter, the estimated channel parameters are extremely unreliable for equalization with conventional adaptive methods. We approach the channel acquisition, tracking and equalization problems jointly, and exploit the richness of the inherent structural relationship between the channel parameters and the data sequence by repeated use of available data through a forward- backward optimization procedure. This enables the fuller exploitation of the available data. Our simulation studies show that significant performance gains are achieved over conventional methods. In the final part of this thesis, we address the problem identifying and equalizing multi path communication channels in the presence of strong CCl. By considering CCI as stochasic processes, we find that temporal diversity can be gained by observing the channel outputs from a tapped delay line. Together with the assertion that the finite alphabet property of the information sequences can offer additional information about the channel parameters and the noise-plus-covariance matrix, we develop a spatial temporal algorithm, iterative reweighting alternating minimization, to estimate the channel parameters and information sequence in a weighted least squares framework. The proposed algorithm is robust as it does not require knowledge of the number of CCI nor their structural information. Simulation studies demonstrate its efficacy over many reported methods.

In this dissertation, channel estimation and channel prediction are studied for wireless communication systems. Wireless communication for time-variant channels becomes more important by the fast development of intelligent transportation systems which motivates us to propose a reduced rank channel estimator for time-variant frequency-selective high-speed railway (HSR) systems and a reduced rank channel predictor for fast time-variant flat fading channels. Moreover, the potential availability of large bandwidth channels at mm-wave frequencies and the small wavelength of the mm-waves, offer the mm-wave massive multiple-input multiple-output (MIMO) communication as a promising technology for 5G cellular networks. The high fabrication cost and power consumption of the radio frequency (RF) units at mm-wave frequencies motivates us to propose a low-power hybrid channel estimator for mm-wave MIMO orthogonal frequency-division multiplexing (OFDM) systems. The work on HSR channel estimation takes advantage of the channel's restriction to low dimensional subspaces due to the time, frequency and spatial correlation of the channel and presents a low complexity linear minimum mean square error (LMMSE) estimator for MIMO-OFDM HSR channels. The channel estimator utilizes a four-dimensional (4D) basis expansion channel model obtained from band-limited generalized discrete prolate spheroidal (GDPS) sequences. Exploiting the channel's band-limitation property, the proposed channel estimator outperforms the conventional interpolation based least square (LS) and MMSE estimators in terms of estimation accuracy and computational complexity, respectively. Simulation results demonstrate the robust performance of the proposed estimator for different delay, Doppler and angular spreads. Channel state information (CSI) is required at the transmitter for improving the performance gain of the spatial multiplexing MIMO systems through linear precoding. In order to avoid the high data rate feedback lines, which are required in fast time-variant channels for updating the transmitter with the rapidly changing CSI, a subframe-wise channel tracking scheme is presented. The proposed channel predictor is based on an assumed DPS basis expansion model (DPS-BEM) for exploiting the variation of the channel coefficients inside each sub-frame and an autoregressive (AR) model of the basis coefficients over each transmitted frame. The proposed predictor properly exploits the channel's restriction to low dimensional subspaces for reducing the prediction error and the computational complexity. Simulation results demonstrate that the proposed channel predictor out-performs the DPS based minimum energy (ME) predictor for different ranges of normalized Doppler frequencies and has better performance than the conventional Wiener predictor for slower time-variant channels and almost the similar performance to it for very fast time-variant channels with the reduced amount of computational complexity. The work on the hybrid mm-wave channel estimator considers the sparse nature of the mm-wave channel in angular domain and leverages the compressed sensing (CS) tools for recovering the angular support of the MIMO-OFDM mm-wave channel. The angular channel is treated in a continuous framework which resolves the limited angular resolution of the discrete sparse channel models used in the previous CS based channel estimators. The power leakage problem is also addressed by modeling the continuous angular channel as a multi-band signal with the bandwidth of each sub-band being proportional to the amount of power leakage. The RF combiner is designed to be implemented using a network of low-power switches for antenna subset selection based on a multi-coset sampling pattern. Simulation results validate the effectiveness of the proposed hybrid channel estimator both in terms of the estimation accuracy and the RF power consumption. / Graduate

Fundamental to the understanding of how rivers transform and shape our environment is the role of turbulence and complex, three-dimensional flow, such as vorticity, in sediment transport and erosion. However, classical sediment transport models (or formulae) are based, for the most part, on boundary shear stress and do not incorporate these natural phenomena. This is understandable given that the relationship between turbulence generation, intensity and form; the influence of turbulence on mobile sediment; and the magnitude and patterns of deposition and erosion are complex and difficult to quantify. Nevertheless, the failure to incorporate turbulence and complex, three-dimensional flow into existing models means that our understanding of sediment transport remains incomplete. The central hypothesis of this thesis is that vortical motion is the main factor in the development of local scour. To examine this, a comprehensive experimental approach was taken. This approach involved detailed measurements of the unique three-dimensional patterns of velocity, turbulence and bed morphology associated with flow in an open channel with and without the presence of submerged groyne-like structures (stream barbs and wing dikes) at both the laboratory and field scales. Using densely spaced velocity data and new techniques for processing and analyzing these data, it was possible to identify complex features of the flow field at both scales, such as the presence of vortex cores, and correlate these with changes in bed levels. These data provide a unique perspective of the spatial variability of velocity, turbulence and bed morphology in open channels (and in particular in channel bends, where flow is highly three-dimensional). In particular, it was found that, of the flow parameters analyzed, turbulent stresses were most correlated to scour in channel bends without the presence of structures. In contrast, vorticity was found to be the dominant factor in local scour hole development in the vicinity of submerged structures at both the laboratory and field scale. Such data are rare (if non-existent) in existing experimental research, particularly at the scale of a large natural river. This research also examines stream barb design and performance. Stream barbs (also known as submerged groynes or spur dikes) are a relatively novel approach to stream bank protection and are not common in Canada. The design and installation of stream barbs in a creek in Ottawa (Sawmill Creek) was undertaken to serve as a demonstration project for the use of these structures in a semi-alluvial channel, for which no such case studies exist. Three years of monitoring the site and the performance of these structures to reduce bank erosion and improve aquatic habitat have been carried out. Laboratory testing of stream barb performance was also undertaken and showed that with improper design, the outer bank in a channel bend may be more susceptible to erosion due to excessive local scour downstream of the barb. In particular, to avoid bank erosion downstream of the barb, barbs should be small, such that they create minimum flow obstruction, and include a bank key that is wider than the barb itself and extends in the downstream direction.

Recently, it has been postulated that fatigue is a mechanism to protect the muscle fiber from deleterious ATP depletion and cell death. The ATP-sensitive potassium (KATP) channel is believed to play a major role in this mechanism. Under metabolic stress, the channels open, reducing membrane excitability, Ca2+ release and force production. This alleviates energy demand within the fiber, as activation of the channel reduces ATP consumption from cellular ATPases. Loss of KATP channel activity during fatigue results in excessive intracellular Ca2+ ([Ca2+]i) levels, likely entering the fiber through L-type Ca2+ channels. It has been demonstrated that when mouse muscle lacking functional KATP channels are stimulated to fatigue, ATP levels become significantly lower than wild type levels. Thus, it was hypothesized that a lack of KATP channel activity impairs energy metabolism, resulting in insufficient ATP production. The focus of work for this M.Sc. project was to test this hypothesis. Fatigue was elicited in Kir6.2-/- FDB muscles for three min followed by 15 min recovery. After 60 sec, a 2.6-fold greater glycogen breakdown was observed in Kir6.2-/- FDB compared to wild type FDB. However, this effect disappeared thereafter, as there were no longer any differences between wild type and Kir6.2-/- FDB in glycogen breakdown by 180 sec. Glucose oxidation after 60 sec was also greater in Kir6.2-/- FDB compared to wild type FDB. However, levels of oxidation failed to increase in Kir6.2-/- FDB from 60 to 180 sec. Calculated ATP production during the fatigue period was 2.7-times greater in Kir6.2-/- FDB, yet measured ATP levels during fatigue are much lower in Kir6.2-/- FDB compared to wild type FDB. Taken together, it appears that muscle energy metabolism is impaired in the absence KATP channel activity.

Speech intelligibility represents how comprehensible a speech is. It is more important than speech quality in some applications. Single channel speech intelligibility enhancement is much more difficult than multi-channel intelligibility enhancement. It has recently been reported that training-based single channel speech intelligibility enhancement algorithms perform better than Signal to Noise Ratio (SNR) based algorithm. In this thesis, a training-based Deep Neural Network (DNN) is used to improve single channel speech intelligibility. To increase the performance of the DNN, the Multi-Resolution Cochlea Gram (MRCG) feature set is used as the input of the DNN. MATLAB objective test results show that the MRCG-DNN approach is more robust than a Gaussian Mixture Model (GMM) approach. The MRCG-DNN also works better than other DNN training algorithms. Various conditions such as different speakers, different noise conditions and reverberation were tested in the thesis.

In a telemetry system, it has been frequently proved that multipath channels and transmission noise are the most critical sources of distortion. While equalization allows a strong limitation of the multipath effects, the noise impact can be efficiently reduced if forward error correction is used. This paper proves that the combination of blind equalization and a powerful FEC like LDPC strongly improves bit error rates for the SOQPSK modulation. We also prove that a LDPC code is able to fully correct the residual errors that may persist at the equalizer output. In other terms, the combination of equalization and LDPC code enables quasi-error free transmissions in various channel scenarios that represent the various phases of a telemetry mission.

Degradation in visibility is often introduced to images captured in poor weather conditions, such as fog or haze. In this paper, a fusion-based transmission estimation method is introduced to adaptively combine two different transmission models. Specifically, the new fusion weighting scheme and the atmospheric light computed from the Gaussian-based dark channel method improves the estimation of the locations of the light sources. To reduce the flickering effect introduced during the process of frame-based dehazing, a flicker-free module is formulated to alleviate the impacts. The system assessments show this approach is capable of superior defogging and dehazing performance, compared to the state-of-the-art methods, both quantitatively and qualitatively

The ATP-gated P2X7 receptors (P2X7Rs) play a key role in the release of pro-inflammatory cytokines in response to immunological challenges. Pannexin 1 (Panx1), conventionally described as a hemichannel forming protein, was suggested to be involved in the formation of the P2X7 large pore, which provides a conduit for large molecules such as fluorescent dyes. Firstly, this thesis demonstrated that the P2X7R-mediated dye uptake, a phenomenon attributed to the activation of Panx1, was suppressed by acidic pH and this inhibition was abolished in a P2X7 mutant (aspartic acid 197 to alanine) that was insensitive to extracellular pH. Then, the functional properties of human or mouse Panx1 in HEK293 cells were analysed in the absence of P2X7. The Panx1 currents were not affected by extracellular/intracellular calcium, but were reversibly inhibited by adenosine triphosphate (ATP) and non-specific anion channel blockers. Ion substitution experiments showed that Panx1 was permeable only to monovalent anions and single channel studies revealed a medium sized unitary conductance of Panx1 (~65 pS). Based on the evidence, this thesis concluded that Panx1 is an anion channel but not a hemichannel as originally proposed.

Steel tubular structures are becoming more widely used in recent years because of their excellent structural performance as well as appearance. For structures with steel hollow section (SHS) columns and concrete filled tubular (CFT) columns, several types of joint may be used to connect to wide flange beams. The topic of this research is a relatively new type of joint, the 'reverse channel' connection, in which the legs of a channel are welded to the tube thus converting the closed tubular column into an open section for connection purpose. However, since tubular columns have only recently become more popular in frames, there is a lack of research to quantify the moment-rotation characteristics of connections between I beams and tubular columns. A total of eight tests on loaded steel beam to RHS column connections using reverse channel and endplate, up to collapse, were carried out. The objectives of these tests were to investigate the effects of different connection details on connection performance and to provide experimental data for validation of numerical simulations. Validation of numerical modelling was carried out to check accuracy of the numerical simulation model developed using the general finite element package ABAQUS. The validated ABAQUS model was then used to perform various parametric studies to provide insight into structural behaviour. A systematic approach to quantify the load carrying capacity of the reverse channel web and the column face were adopted based on yield line theory. This thesis presents the derivation and validation of the analytical approach. The moment resistance of the whole connection can be obtained by assembling all the components together. A procedure for calculation was proposed to provide feasibility for designer and the analytical results were compared with the test results.