<p>Time frequency representations provide us with an efficient way to process non-stationary signals and systems with time-dependent spectrum. In this thesis, a tutorial review of several time-frequency representations is presented. The historical development and the properties are discussed with the emphasis being placed on the Wigner distribution (WD). Three important applications of time-frequency representations are studied. They are the tracking of instantaneous frequency, high resolution estimation of frequencies in the non-stationary environment, and the design of signals for optimum simultaneous estimation of time delay and Doppler shift. An efficient way to calculate the Cramer-Rao lower bound for the time-varying frequency estimation is obtained. The theoretical performance of the WD when applied to the estimation of this time-varying frequency is derived and the simulation examples are provided. Using the properties of the instantaneous spectrum, new algorithms are developed which can accurately estimate the frequencies of the non-stationary signals with fast varying amplitude. A particular example of the applications is the frequency estimation in Search and Rescue Satellite (SARSAT). For tests on real SARSAT signals, our new algorithms exhibit superior performance to other spectrum estimation methods. The Cramer-Rao bound (CRB) for the joint time delay and Doppler shift estimation is derived and it is proved that the joint estimates of these two parameters using time-frequency representations are optimum under high SNR in the sense that it is unbiased and the variance reaches the CRB. Several efficient methods are developed to obtain the optimum signal for the joint time delay and Doppler shift estimation under different practical constraints.</p> / Doctor of Philosophy (PhD)

<p>This thesis investigates the problem of adaptive control for a class of nonlinear systems which are explicitly linearizable by nonlinear state feedback. A theoretical framework and a systematic design procedure have been established for adaptive control of feedback linearizable systems with parametric and dynamic uncertainties. An error model for adaptive tracking problem is introduced for the first time considering both the parametric and dynamic uncertainties. The significance of the error model lies its explicit physical meanings. It can serve as a basis for the development of adaptive strategies and control algorithms for feedback linearizable systems. Four new adaptive algorithms have been proposed. The stability and parameter convergence of these algorithms are theoretically established by using two different approaches. The robustness of the algorithms for adaptive regulation problem has been analyzed. The tracking ability and effect of initial parameter estimates have been studied also. The restrictive matching conditions and nonlinearity growth conditions are two of the main problems appearing in literature of adaptive control for nonlinear systems. The two problems have been solved here for the first time. A Model Reference Adaptive Control algorithm and an Augmented Error Control algorithm have been presented to remove these restrictive limitations. The class of nonlinear systems for which adaptive control can be applied has been substantially enlarged. A comparison between adaptive control schemes and nonadaptive control schemes has been made. The results of comparison show that the performance of adaptive controllers is superior to that of nonadaptive state feedback controllers. A decentralized adaptive control strategy is introduced for nonlinear systems. It has been shown that a decentralized adaptive control system is much easier to implement than a centralized control system and is more robust to structural disturbance. To demonstrate practical applications, the proposed adaptive control algorithms have been applied to the control problems of a class of robotic manipulators. Three different types of adaptive tracking problems of manipulator systems have been investigated. The results of simulations demonstrate considerable improvements in tracking accuracy over the traditional inverse dynamics control methods in the presence of significant parametric uncertainty.</p> / Doctor of Philosophy (PhD)

<p>The use of ground probing radar for the nondestructive evaluation of structures has been in use for more than a decade. Through the examination of the subsurface reflection waveforms, determinations may be made regarding the profile geometry as well as detecting the presence of structural faults undetectable from surface inspections. Although automated systems have been previously created, the arduous task of interpretation of the large volumes of data produced by these radar surveys are still best handled by human experts. The research presented in this dissertation details the development of novel techniques which allow an Artificial Intelligence based system to demonstrate comprehensive structural and fault analysis capabilities. Development of an autonomously defined and system trained neural network based radar waveform filter allows reflection event feature extraction. The three-dimensional concept of the bridge structures and the system's ability to accurately model the diverse cross-sections are combined with the neural preprocessed data to allow data abstraction. The positional relationship present in bridge radar surveys are exploited to develop a unique spatiotemporal foundation for analysis. The comprehensive expert system was tested with real bridge radar data and the system results correlate well with human expert analysis.</p> / Doctor of Philosophy (PhD)

<p>This thesis addresses physics-based microwave device modeling and circuit optimization, including conventional and statistical device modeling, performance-driven and yield-driven circuit design. Approaches for physics-based device modeling are reviewed. Fundamental techniques of physics-based analytical MESFET modeling are presented. Device performance and parameter extraction with physics-based models (PBMs) are discussed. Nonlinear circuit analysis with PBMs integrated into the harmonic balance (HB) method is presented. A detailed formulation of the HB equations with MESFET PBMs is given. An efficient Newton method for solving the HB equations is discussed. Gradient-based optimization for circuit design is addressed. Physics-based circuit optimization integrates efficient adjoint sensitivity analysis approaches, the HB simulation method and PBMs. The physical (geometrical, material and process-related) parameters can be directly treated as design variables. Simultaneous device-circuit design is facilitated. The features of physics-based circuit optimization are demonstrated by two circuit design examples. Statistical modeling at different levels is discussed. Statistical parameter extraction and postprocessing are used to obtain statistical models to predict parameter statistics. The resulting statistical device models are verified by comparing the statistics of measurements with the corresponding statistics obtained by Monte Carlo simulation. Statistical modeling with equivalent circuit models (ECMs) and PBMs is demonstrated. Yield-driven circuit design is addressed based on a one-sided ℓ₁ optimization algorithm with a generalized ℓｐ function. Yield optimization of MMICs with PBMs for passive and active devices is discussed. Its features are demonstrated by a three stage X-band MMIC amplifier design. A comprehensive approach to predictable yield-driven circuit design exploiting a novel statistical model is presented. For the first time, the yield estimated by Monte Carlo simulation is shown to be consistent with the yield predicted directly from device measurement data. Simultaneous device-circuit yield optimization assisted by yield sensitivity analysis is also demonstrated.</p> / Doctor of Philosophy (PhD)

This thesis has been directed toward the problem of deriving a computable tight lower bound on the error of DOA estimation with the array processing. This work is developed based on the logic implied in Ziv-Zakai's idea and the work for Cramer-Rao lower bound (CRLB). A profound understanding of Ziv-Zakai's idea is presented. The lowest bound on the variance of DOA estimate with one incoming signal is derived applying the logic of Ziv and Zakai. Then the modified Ziv-Zakai lower bound (MZLB) on the covariance matrix of the multiple DOA estimates is developed. The theoretic analysis and the simulation results show that MZLB is a tight lower bound over a wide range of signal-noise ratio. It follows the SNR-threshold phenomenon occurring in the performance of the DOA estimation well, and it is easily computable. It is proved that, the maximum-likelihood estimation of DOA parameters based on Data Model(2) discussed in this thesis is asymptotically efficient. / Doctor of Philosophy (PhD)

<p>Channel sharing by orthogonal separation is fundamental to the designs of an extremely wide variety of multiple-user communications systems. Popular examples in digital communications include TDMA, FDMA, CDMA, and the temporary channel that exists during successful transmission in random-access channels. These engineering solutions have been so successful in practice that it may not be recognized that they were limited by the existing technology. Orthogonality is not, in general, a criterion of optimality. For example, in the additive multiple-user AWGN channel, performance is largely determined by the minimum Euclidean distance between signals at the receiver. Yet, there appears to be no signal design work based on maximizing the minimum Euclidean distance in the overall summed set of signals. The symbol-synchronous linearly-additive multi-user channel is considered in the first part of this work. Rules are presented to realize multi-user coded signals that do not lose minimum Euclidean distance in the receive space. In particular, superposition onto BPSK streams with unit symbol-energy is considered. A trellis code based on 4-dimensional unit-energy signals is constructed which allows a second user to transmit. 3 bits every 4 BPSK intervals without reducing the minimum Euclidean distance in the superposed signal. This design obtains a greater summed rate than previous two-user codes. Furthermore. less energy per bit is required for a given minimum distance. The collision event of the random-access channel is also reconsidered. It is shown that overlapping collisions are often separable by the proper processing algorithm. A generalization of ALOHA is given where 'slot'-collisions are replaced by 'subslot'-collisions, where, in general, there is more than one subslot per slot, with one subslot being equivalent to slotted-ALOHA. Throughputs obtained depend on the complexity available. For example, relative to a single user continuously transmitting, 91% throughput is achieved with 4 subslots, which requires an 8-state Verdu-Viterbi trellis detection algorithm, while 159% throughput is achieved with 10 subslots. requiring 512 states. Potentially, systems with K-subslots obtain K carrier-sense channels when carrier-sensing is available for each subslot. Although it was not the goal of this work, some results may have application to single-user communications. For example, the trellis-coded component of the multi-user constructions may be useful for low-rate, low-SNR signalling. Less anticipated is the result that, for certain values of K, a single-user may send K binary symbols separated in time by T/K, using raised-cosine time-domain pulses limited to duration T, with higher spectral-efficiency than M = 2ᵏ level PAM signalling with raised-cosine frequency-domain pulses at rate 1/T. This particular result appears to be unknown, nevertheless it is less relevant today in light of the more substantial gains that have been achieved with (single-user) trellis-coded modulations. Future investigations and other possible directions of research stemming from this work are discussed throughout and in the concluding chapter.</p> / Doctor of Philosophy (PhD)

<p>A localization system based on the Institute of Electrical and Electronics Engineers 802.11™ wireless local area network standard enables cost-effective localization measurements. It also integrates data dissemination within the system, unlike the popular NAVSTAR global positioning system. This thesis addresses two significant deficiencies in such a system: potentially low localization accuracy and high power usage of the communication infrastructure. A design for a low-cost solar powered localization augmentation node (SPLAN) is proposed to provide accuracy-increasing localization assistance service to localization tags-small battery powered units equipped with an IEEE 802.11 transceiver. Power saving network protocols for the SPLAN/ tag architecture are developed to extend the tag battery lifetime and allow the SPLAN to be cost-effectively supplied by renewable energy sources such as solar energy.<br />A comprehensive power consumption model is developed for two complementary communication systems: plain tag and smart tag. Through its use, significant SPLAN energy conservation for low tag densities was demonstrated in the plain tag system, while preserving legacy compatibility. In the smart tag system even greater energy gains were obtained for both the SPLAN and the smart tag, up to maximum supportable tag densities. This was possible through the use of a custom communication protocol designed specifically for smart tag system power saving. Two innovative aspects of the protocol include timing-assisted scanning of management beacon packets and an optimized procedure for dissemination of raw localization data.</p> / Master of Applied Science (MASc)

<p>This thesis is concerned with the removal or reduction of noises in high resolution<br />video sequences. Many video denoising techniques have been published in the past<br />two decades) with or without motion compensation. They vary in a wide range of<br />complexity) performance, and implementation cost. Also) many existing video denoisers make simplistic assumptions on noise statistics and motion type, and hence<br />their performance depends on the validity of the assumed noise and motion models.<br />To improve the performance and robustness of existing methods) we propose a new<br />joint spatial-temporal video denoising algorithm that combines multihypothesis interframe motion compensation and directional intra-frame filtering. The algorithm takes into account general compound motions, including both global camera motion and<br />individual object motion(s). An affine motion model is used to characterize the global<br />camera movement, whereas a blockwise translational motion model is used to approximate local object motions. Quadtree data structure is used to organize and speed up the computations of block-based motion estimation. Quadtree-structured diamond search is conducted so that a large area can be examined in motion estimation at a low computational cost.<br />In order to achieve the best possible visual quality we augment motion-compensated<br />temporal interframe denoising operation by an intra-frame denoising operation of<br />adaptive directional filtering. The directional filter is designed for the local signal<br />waveform and noise level, and it has the advantage of effectively suppressing noises<br />without blurring edges.<br />The proposed video denoising algorithm is implemented and tested extensively on<br />high-resolution digital cinema contents. The experimental results demonstrate the<br />competitive advantages of the new algorithm in both visual quality and processing<br />throughput.</p> / Master of Applied Science (MASc)

<p>In past years, the evolution in communication technology has led to a need for<br />highly-integrated, low-power, and low-cost circuit designs for wireless applications.<br />The demand for radio frequency (RF) wireless transceiver operating at 2.4GHz<br />band has attracted considerable research interest. The performance of such<br />transceivers depends heavily on that of each of the individual blocks such as lownoise<br />amplifiers and mixers. However, there are very few designs that focus on<br />connecting the single-ended output low-noise amplifier (LNA) to a double-balanced<br />mixer without the use of on-chip transformer. This kind of receiver front-end is designed<br />to achieve high integration and low power consumption.<br />In recent years, Ultra-Wideband (UWB) technology has developed very rapidly<br />due to its high data transmitting rate and low power consumption. Meanwhile, the<br />design of an Ultra-Wideband low-noise amplifier (LNA) has become an important<br />challenge since it is normally the front-end of the radio frequency (RF) receiver system.<br />Low power consumption of an UWB LNA is a critical requirement for UWB<br />application such as portable devices or especially in biomedical systems. However,<br />the design should not only focus on low power, but also focus on optimizing other<br />performances at competitive levels over the entire bandwidth, where the Federal<br />Communication Commission (FCC) has allocated 7.5 GHz of bandwidth from 3.1GHz to 10.6 GHz for Ultra-Wideband.<br />This thesis focuses on the design of a fully-integrated RF receiver front-end including<br />a narrow-band LNA followed by a double balanced mixer. The receiver operates at 2.4 GHz and produces an output signal at 300 MHz. The circuit is designed and fabricated using TSMC O.lS-p,mCMOS technology. In order to translate the single-ended RF output signal from the LNA into the differential input pair of the mixer, a main novel idea of this design is to use one PMOS and one NMOS instead of two NMOS devices for the RF amplification stage of the double balanced mixer. The circuit achieves 16.3dB gain and 6.74mW power consumption while using 2.08 mm2 chip area.<br />Another design presented in this thesis is a UWB LNA with special emphasize<br />on low power consumption as well as on optimizing the overall performance. The<br />circuit is designed using TSMC 0.13-p,m CMOS technology. It achieves a very flat<br />gain of 10.3-12.1 dB and 3.4-5.9 dB noise figure (NF) throughout the entire bandwidth<br />of 3.1-10.6 GHz. The power consumption is 2.81 mWwhich is extremely low compared to other designs and the chip area is 0.48mm2 • The overall performance is also competitive according to its figure of merit (FoM).</p> / Master of Applied Science (MASc)

<p>Free-space optical (FSO) links are competitive wireless links offering high data rate,<br />security and low system complexity. Compared to radio-frequency (RF) links, FSO<br />links offer high rates at gigabit-per-second (Gbps) level and relatively low cost. However, atmospheric scintillation and misalignment between optical transmitter and receiver impair data rates of FSO links. Scintillation and misalignment are slow fading<br />processes with a fading interval of 10's ms. Conventional fixed-length channel coding<br />which interleave data blocks are unrealistic to overcome this slow fading due to block<br />length of 10's of megabits. Also, because of the Gbps data rate, data rate adaptation<br />to channel conditions are expensive.<br />In this work, short-length (16 - 1024) Raptor codes are designed to overcome the<br />slow fading of FSO channels. These Raptor codes are applied at the packet-level with<br />high data rate and low decoding complexity. The Raptor encoder and decoder can be<br />easily implemented in any software or hardware form. The practicality of these Raptor<br />codes is demonstrated by a Raptor encoder and decoder which are implemented in<br />field-programmable gate array (FPGA) and shown to support a 1.22 Gbps encoding<br />and 714 Mbps decoding rate with a 97 mW low power consumption and 26360 gate<br />circuit scale. High-speed transmission at Gbps level is easily satisfied by the same<br />design implemented in an application-specific integrated circuit (ASIC).</p> <p>Two applications of these short-length Raptor codes in FSO links are presented<br />in this work. Firstly, these Raptor codes are applied to hybrid FSO/RF links to<br />achieve high data rate by sending Raptor encoded packets simultaneously over the<br />FSO and RF links which we term such links as Raptor-coded soft-switching hybrid<br />FSO/RF links. The performance of these Raptor codes in the hybrid FSO/RF links<br />is simulated in a realistic channel model based on climate data of three Canadian<br />cities. For a 1 Gbps FSO link combined with a 96 Mbps WiMAX RF link, the softswitching<br />system achieves an average rate of 472 Mbps using the implemented Raptor<br />code while hard-switching technique achieves only 112 Mbps on average.<br />Secondly, these Raptor codes are applied in mobile FSO links for an unmanned<br />aerial vehicle (DAV). This mobile FSO link suffers from severe instantaneous misalignment. For packet-level transmission, the time varying misalignment is unknown<br />to the transmitter and causes data packet corruption and erasure. As a result, the<br />application of conventional fixed-rate erasure coding techniques is difficult. In this<br />work, short-length Raptor codes are applied in such mobile FSO channels. A key advantage of Raptor codes is their independence on channel state, no matter how large<br />the misalignment. With a 1 Gbps transmitter, the designed Raptor code with k = 64<br />message packets offers 650 Mbps data rate when transmitting power is 20 dEmo In<br />contrast, a traditional automatic repeat-request (ARQ) algorithm technique on the<br />same FSO jitter channel achieves a rate of 70 JVlbps.</p> / Master of Applied Science (MASc)