<p>This research investigates the development of a methodology for designing self-tuning feedback linearizing control laws. If the conventional architecture for linear plant self-tuning systems is applied to the feedback linearization case, it is shown that the estimation algorithm gradient vanishes as the parameter estimates approach the true parameter values. Vanishing of the gradient causes the covariance matrix to increase without bound and consequently system failure. A new architecture is presented that eliminates the covariance problem but does not yield a direct estimation of the nonlinear plant parameters. The parameters estimated in the new architecture are composites of the true parameter values of the nonlinear plant and their estimated values. An adaptive law is designed to interpret an error equation formed from the composite parameters and asymptotically converge to the true nonlinear plant parameter values. A stability proof and convergence properties for the adaptive law are given. Sufficient conditions for a nonlinear plant to be capable of self-tuning in the new architecture are specified. The new method is demonstrated with simulations of an arbitrarily chosen nonlinear plant and two plants of practical interest. One plant is a chemical reactor running an exothermic process where reactor temperature is the controlled variable. The other plant is a bioreactor where control of the substrate concentration in an anaerobic digestion process is the objective. In both cases the method developed in this thesis offers performance improvements as compared with previously published results on control of these processes by other methods.</p> / Doctor of Philosophy (PhD)

<p>The purpose of this thesis is to design bandwidth efficient codes for use in a time selective fading channel. A time selective fading channel, the correlated Rayleigh fading channel, is analyzed and criteria for the design of bandwidth efficient codes are presented. These criteria are used to design new codes which are inherently resistant to Rayleigh fading. The new codes are analyzed and a decoding algorithm for the codes is presented. Simulations of the performance of the codes and the decoding algorithm are then presented. Simulations of the performance of the codes and the decoding algorithm are then presented. Finally, a channel estimator that generates the necessary information for decoding is presented, analyzed, and simulated.</p> / Doctor of Philosophy (PhD)

<p>This thesis presents an in-depth investigation of new broadband millimeter wave antennas, which can be used in millimeter wave point-to-point and point-to-multipoint communications. A very powerful design tool--the Finite Difference Time Domain (FDTD) method is presented in its regular uniform grid form, which is referred to as Yee's method, its non-uniform grid form, and its General Curvilinear Coordinate (GCC) form. Accompanying formulas are presented in each case. The advantages of these different formulas are discussed, their use for solving different complex electromagnetic problems is demonstrated, as well. Different absorbing boundary conditions (ABC) are used to design antennas, their beam forming networks and transitions, for example Mur's first order ABC, Litva's dispersive ABC and Modified Perfect Matched Layers (M-PML). Here Litva's ABC is presented in detail, and some case studies are given to prove that it is a very good ABC for printed circuit applications. A research based FDTD 3-Dimensional Structure Simulator (FDTD 3D SS) is used to design these new wide-band antennas and transitions. More than 500 simulations have been calculated for this study within less than a year. The Tapered Slot Antennas (TSA) and different type feed structures for TSA are reviewed in Chapter 2. As a member of the class of traveling wave antennas, the TSA's are wide-band elements with very good directivities. Based on the Suspended Microstrip Line (SML) and the Inverted Microstrip Line (IML), three new feed structures have been developed, by using FDTD 3D SS, for millimeter wave wide-band applications. The Vivaldi antenna, which belongs to the traveling wave antenna category and shares the same features as TSA, is used as a very short element (antenna length L < λo /2) to design an 120-degree-vertical-polarization sector antenna. The experimental study shows that the design gives very good results and meets the specifications in real applications. It has a potential to be used as a sector antenna with different horizontal beam-widths. This design shows antenna engineers a novel approach to design sector antennas without worrying about the antenna input impedance bandwidth. (Abstract shortened by UMI.)</p> / Doctor of Philosophy (PhD)

<p>When an aircraft or marine vessel is in distress, the emergency unit is activated either automatically or by manually setting a switch. The signal is then received by an orbiting satellite as it sweeps out a path over the search and rescue satellite aided tracking (SARSAT) region of interest. Due to the relative motion between the satellite and the Emergency Locator Transmitter (ELT) or the Emergency Position Indicating Radio Beacon (EPIRB) unit, the signal received at the spacecraft is Doppler shifted. New processing techniques for detecting ELT signals in an interfering background are investigated, based on not only computer simulations but also experimental data. The new methods involve the use of the Geometrically Averaged Spectrum (GAS) and the Harmonically Averaged Spectrum (HAS) in addition to the Arithmetically Averaged Spectrum (AAS) adopted by the existing system. Specifically, this thesis investigates the theoretical performance of AAS, GAS and HAS averaging in practical circumstances involving signals, noise and interference. It is shown that the carrier peaks of good ELT signals can be substantially enhanced in an interfering background. The treatise identifies certain types of interference such as Carrier interference with strong or weak AM, On-Off Modulation interference, Swept Carrier interference, and Ringing Response interference. Using computer simulation, it is demonstrated that under certain conditions the effects of background interference can be greatly reduced and the detection of ELT carrier peaks can be improved by employing a combination of AAS, GAS and HAS averaging. Real SARSAT pass data signals are fed into the computer and processed using AAS, GAS and HAS averaging. The background interference postulated in this thesis is found to exist and real ELT signals are identified. ELT signals representing the NIC model and a new model, called the Frequency Modulated ELT (FME) model, are identified along with a new ELT signal for which a model cannot presently be envisaged. The results of letting AAS, GAS, and HAS work together are compared to that of using only AAS. It is seen that the new strategy shows superior performance over the original method in a congested signal environment even though interference power exceeds ELT signal power. (Abstract shortened by UMI.)</p> / Doctor of Philosophy (PhD)

<p>Component-Oriented Software Technology started to emerge during the last few years. The purpose is to make it easier to build new applications from existing components that can be obtained from Component-Oriented Toolkits (CO-Toolkits for short). Experience shows that CO-Toolkits are hard to understand and use. Therefore, developers would write their own code from scratch using programming languages that they know instead of using components from such CO-Toolkits. In this research, we have used ADAPTIVE Communication Environment (ACE for short)--as an example of CO-Toolkits--to establish a set of initial assumptions for good design practice from the perspective of their file structures, design and implementation, and documentation of CO-Toolkits. Then (i) we show how to formalize rules for good file structures, design and implementation, and documentation that are based on these initial assumptions, (ii) we show that checking tools can be built based on these formalized rules, and then (iii) we demonstrate that such tools can be used to check existing CO-Toolkits code and report any rule violations back to developers.</p> / Doctor of Philosophy (PhD)

<p>An output coefficient extracted from motor design sizing calculations is cast into a form whereby it can be implemented as an actuation independent measure of performance. The way in which this measure scales for both electrically actuated motors and magnetically actuated motors is investigated. An engineering process for judging the advantages and disadvantages of each actuation scheme as it applies to particular motor designs is discussed. The performance capabilities of an electrically actuated variable capacitance motor are compared to those of a magnetically actuated alternative using the output coefficient as a comparative measure. A transition point is estimated which represents the scale at which the benefits of electrical actuation over magnetic actuation put electrically actuated devices in the realm of practical consideration. The two axis or dq0 method of analysing conventional magnetically actuated machines is extended and applied to the topology of an electrically actuated variable capacitance silicon micromotor. A "conventional" finite element technique is used to obtain estimates for the direct axis and quadrature axis synchronous capacitive susceptances. The energy based formulation employed eliminates the need to extract self and mutual phase capacitances and produces the desired axis capacitances directly. This novel method brings attention to higher order harmonics in the direct axis and quadrature axis charge waveforms. A physical explanation based on positive, negative and zero sequence phase charge is derived accounting for the observed rotor position-dependent characteristics. A dual bounds approach is applied to a typical two-dimensional micromotor structure. The direct axis capacitance is computed for the conditions of maximum stator pole-rotor pole overlap with the excitation EMT placed in two "extreme" conditions. The first condition is placement along the rotor's direct axis and the second condition is placement along the rotor's quadrature axis. Quadrilaterals are used to discretize a ninety degree (electrical) section of a silicon micromotor structure with a stator pole to rotor pole ratio of three to two. Two separate discretizations are considered, one for each of the stator EMF conditions previously noted. A method in which the same quadrilateral discretization can be used for both the upper and lower capacitance bound is presented. The notion of a stationary solution for the method of dual bounds is shown to exist. A simple two-dimensional example is given in order to illustrate how the linear equations are derived and to bring to out the ability to control the coupling of the resultant linear equations. Substantial computational savings, especially in the three-dimensional case can be realized (when compared to the finite element method) and warrant future examination and application in both electrically actuated and magnetically actuated problems. The encouraging two-dimensional results lead to a full three-dimensional extension of the theory, in particular, the development of novel geometrical constructs necessary for dual bound analysis. Formal methods for determining the placement of equipotential slices and flux tube walls inside fundamental cuboid structures are presented. Actual numerical results are presented for two simple three-dimensional constructs. A discretization process is given that shows how a complete three-dimensional silicon micromotor model can be discretized using the geometrical primitives derived in this thesis.</p> / Doctor of Philosophy (PhD)

<p>The objective of satellite orbit determination is to find accurate values of a set of orbital elements which describes the orbit of the satellite, using observations of the satellite. The extended Kalman filter has been widely used for the estimation of the orbital states. The purpose of this work is to find an alternative approach that would reduce the amount of on-line computation required. A nonlinear observer is proposed for this application. Its stability problem is studied through the second method of Liapunov. The performance of the "nonlinear observer is then evaluated with simulated orbits. AIthough many secondary effects on the dynamics of the satellite have been omitted, the simulation is indicative of results that can be obtained in real situations.</p> / Master of Engineering (ME)

<p>Over the past decade, lattices have been increasingly recognized as an important source of codes for the Gaussian channel. The 8-dimensional Gosset lattice has figured prominently in these new developments because it offers an asymptotic coding gain of 3 dB over conventional pulse amplitude modulation and can be soft decision demodulated/decoded with a reasonable amount of speed. In the present work, we revive a little used definition of the Gosset lattice and show that codes derived from this construction exhibit a null at dc in their baseband spectrum. Such codes are useful as line codes for baseband signalling on channels that do not support a dc spectral component or for bandpass transmission where spectral shaping is required to combat intersymbol interference.</p> <p>Previous applications of lattice codes have been aimed primarily at voiceband data communications. This thesis was motivated by the need to develop a decoder that would make applications in the multi-megabit range of data transmission possible. To achieve this goal, a two-stage approach to demodulation was developed. The first stage makes a fast estimate of the transmitted vector and has the ability to declare an erasure when it knows its estimate is unreliable. This initial erasure declaring stage controls the throughput of the demodulator. Because it is far simpler than a maximum likelihood demodulator, greater speed is achieved. The second stage is provided to correct the occasional occurrence of an erasure and maintain the error performance of the lattice. To complete the decoder structure, we outline a method of lexicographic ordering the signal set that leads to a compact set of decoder look-up tables used to obtain a binary message sequence from each demodulated vector. Finally, we evaluate the effects of quantization on the probabilities of erasure and error and give results from a Monte-Carlo simulation undertaken to verify the demodulators performance.</p> / Doctor of Philosophy (PhD)

<p>Fiber optic communication is only 30 years old, but it has already become one of the technologies of choice for current and future high-speed communications. Unfortunately, the enormous bandwidth available in optical fiber cannot be accessed as a single communication channel. This is because each user station typically connects tot he network through its electronic interface whose speed is limited to the maximum electronic processing speed. However, access to this bandwidth is possible using parallel channel architectures, where each channel operates at rates which are accessible to electronic processing. Currently, wavelength-division-multiplexing (WDM) is being considered for this purpose. In local and metropolitan area networks, WDM may offer increased transmission concurrency compared with conventional single-channel designs, with only modest increases in station complexity. However, because of device limitations the number of available wavelength channels may initially be less than desired. As a result, spatial reuse of wavelength channels may be required to obtain designs which will support a reasonable number of stations. One option in this case is to perform media access across a spatial wavelength cross connect. In such a system, the total capacity may be n times that of a single passive star network, where n is the number of available wavelengths.</p> <p>In this thesis, WDM cross connect networks are considered as a specific category of optical networks, which efficiently reuse the available wavelengths. The proposed networks are classified into two classes: single-hop and dual-hop.</p> <p>In the single-hop case, time slot assignment techniques are considered which can dynamically schedule communications across the wavelength cross connect network. These new scheduling techniques are developed for a high performance dynamic medium access control based on the traffic demands of the network stations. Some simple random access methods are also introduced for the cases where some higher levels of simplicity are sought. Simulation programs are used to compare different scheduling methods. The results also show that the proposed methods perform quite efficiently.</p> <p>The dual-hop class of WDM cross connect networks proposes a shared buffering scheme which simplifies network operation. These networks consist of two stages. In the first, the wavelength tunability of the user stations is used to route packets from a given local optical network (LON) to a destination LON. When packets arrive at a destination LON, they are buffered and transmitted onto the required destination wavelength. There are a number of advantages to this design including the elimination of protocols which would require both dynamic station transmitter and receiver tunability. This design also takes advantage of increasingly available commercial ATM (Asynchronous Transfer Mode) buffer/switch components. Several hybrid electro-optic options are considered. It is also shown that by using a novel optical multichannel buffering scheme, the number of required buffers can be significantly decreased.</p> / Doctor of Philosophy (PhD)

<p>This thesis reports my research work in the area of trellis coded continuous phase frequency shift keying (CPFSK). Previous approaches [1, 2, 3, 4, 5] applied binary convolutional codes to CPFSK to achieve power and bandwidth efficiency. However, the work in [6] and part of this thesis show that no single approach among previous approaches can be outperformed by the others if only binary convolutional codes are considered.</p> <p>A new coding scheme based on cnovolutional codes on the ring of integers modulo-P is shown to be a natural way to apply trellis coding to CPFSK [7]. Recent work has decomposed CPFSK into two parts; a linear encoder with memory, called the continuous phase encoder (CPE), and a memoryless modulator (MM), where the CPE often has a code structure defined over the ring of integers modulo-P. The combination of modulo-P convolutional channel encoder (CE) and the CPE, is a linear modulo-P encoder. Design examples are given for rate 1/2 coded quaternary CPFSK with modulation indices 1/2 and 1/4, and rate-2/3 coded octal CPFSK with modulation index 1/8. Combinations are optimized in the normalized minimum Euclidean distance sense for a given total number of states in the overall maximum likelihood sequence estimation (MLSE) receiver. Numerical results show that this new coding scheme consistently achieves better performance than previous schemes. [1, 2, 3, 4, 5].</p> <p>An upper bound on the bit error probability (BER) for ring convolutionally encoded CPFSK is derived. The bound shows that feedback-free CPFSK usually has a smaller error coefficient than CPFSK. The minimum Eucildean distance is a good parameter for estimating performance, and the ring convolutionally encoded CPFSK has a good BER for both moderate and practical signal to noise ratio.</p> / Doctor of Philosophy (PhD)