<p>In this thesis, we investigate full-wave hybrid Conjugate Gradient - Fast Fourier Transform (CG-FFT) methods for analyzing microstrip arrays. In particular, we investigate spectral preconditioning with the CG-FFT methods.</p> <p>First, a general scheme is introduced for implementing CG-FFT using the spatial discretization. In comparison with other schemes, the proposed one yields greater accuracy and has higher efficiency because no finite difference approximations are involved, all the aliasing and truncation errors are eliminated and the size of the zero padded region is kept to a minimum.</p> <p>A hybrid full-wave CG-FFT method is then developed for analyzing microstrip structures. It combines the proposed CG-FFT scheme and the full-wave complex discrete image technique. With this combination, the spatially discrete scheme is realized which can be used for microstrip structure analysis without losing any full-wave information, while at the same time, measuring only minor computational cost and errors. Therefore the merits of the proposed scheme are extended to microstrip problems straightaway.</p> <p>To further improve the rate of convergence for the CG-FFT method, a new highly efficient spectrally preconditioned CG-FFT method is introduced. This technique takes full ad vantage of the FFT by constructing the special forms of preconditioners and performing the preconditioning in the frequency domain. It uses no any additional memory and only O(NlogN) additional operations for constructing and storing the preconditioner. At the same time it has superior convergence properties compared with the conventional CG-FFT method and other existing preconditioned methods.</p> <p>The hybrid CG-FFT algorithm developed in this thesis is easily interfaced to UNIX simulation software, including a draw editor and an automatic mesh generator. Using this software, we carry out numerical analyses of different types of microstrip antenna arrays. The analyses are corroborated by experimental measurements. A number of array parameters and boundary effects are studied. These are of interest to antenna design engineers, which include: (1) effect of the finite size of arrays, (2) effect of array shape, (3) spurious radiation from the array feed structures, (4) current distributions on the elements, (5) input impedance and (6) array radiation patterns. Conclusion and discussion are addressed. The performance capabilities of the hybrid full-wave CC-FFT method are demonstrated by a modeling study of very large microstrip reflectarrays. Proposals are made for an improved design for the large array, based on the results of the simulations.</p> / Doctor of Philosophy (PhD)

<p>Smart antennas have emerged as one of the key technologies in wireless communications. This thesis focuses on developing algorithms and structures that can be applied to one of the most important types of smart antennas--the multibeam smart antenna--in order to improve the quality and capacity of both existing and future wireless networks. A large database, consisting of vector channel measurement data using an 8-element circular antenna array is analyzed to investigate the underlying characteristics of the multibeam smart antenna in practical propagation environments. Extensive vector channel simulations are also conducted, including the innovative work on the simulation of smart antennas in Frequency Division Duplexing (FDD) systems. Based on the data analysis and simulation results, two multibeam smart antenna algorithms, one for reception and one for transmission, are proposed. Significant performance and capacity improvements can be achieved using these two algorithms. The proposed multibeam smart antenna reception algorithm for DS-CDMA system is expected to have important application in both existing and the third generation CDMA systems, while the proposed multibeam smart antenna transmission algorithm can be used in FDD wireless networks to solve the downlink problem.</p> / Doctor of Philosophy (PhD)

<p>There is an increasing demand today for the deployment of wireless services such as cellular telephony, paging, wireless local area networks, digital broadcast television, and wireless Internet. The demand for bandwidth from these services continues to increase as more people adopt the technology and the applications become more sophisticated. In the recent years, researchers have demonstrated that smart antennas are capable of improving wireless link quality and increasing coverage area. In addition to this, it has also been demonstrated that smart antennas can permit the reuse of the frequency spectrum in the same wireless coverage area using space division multiple access (SDMA). SDMA has the potential to provide a tremendous increase in system capacity over conventional wireless systems. In this thesis we consider the application of smart antennas for packet switched SDMA networks. The system considered is an infrastructure network employing a smart antenna basestation which communicates with portable stations that have omnidirectional antennas. A time division multiple access (TDMA) packet switched system that incorporates SDMA is considered. Several heuristic dynamic slot assignment (DSA) strategies are proposed. DSA is the name given to the process of allocating packets to the time slots in real-time on a frame by frame basis. The primary objective of DSA is to maximize the frame capacity. Both a theoretical Rayleigh fading channel model and experimental data collected using an 8 element circular antenna array built at the Communications Research Laboratory at McMaster University is used to measure the frame capacity of the heuristics. Analytic, simulated, and experimental results demonstrate that the SDMA/TMDA frame capacity is several times higher than the single omnidirectional basestation antenna case. The results also provide insight into how the protocol performance is affected by parameters such as wireless channel model, slot assignment complexity, power control, pedestrian motion in the channel, and signal to noise ratio (SNR). The well known slotted ALOHA protocol which is best suited for bursty data communication between a large population of portable stations is adapted for operation with a smart antenna basestation. Versions of this protocol are considered where contention takes place in the data slot directly and when a reservation channel feeds a contention free data channel. The DSA heuristics developed earlier are applied to the contention free data channel to improve system capacity. An SDMA version of the slotted ALOHA protocol is modified for operation in a multicell situation. Simulation and analytic results show that the single cell slotted ALOHA system can achieve a many-fold increase in system capacity, especially when intelligent slot scheduling is done at the basestation. The sensitivity of the protocol capacity to factors such as the hardware complexity, packet size, SNR, and protocol complexity is examined. An analysis of the multicell slotted ALOHA protocol reveals that it is possible to reuse the frequency in every cell and still achieve a per cell capacity similar to ordinary single-cell slotted ALOHA with only a modest degree of smart antenna beamforming hardware at the basestations. The DSA heuristics are then extended to multicell situations. Several DSA techniques requiring various degrees of coordination between the basestations are examined. DSA enhancements are also proposed for systems with variable length packets.</p> / Doctor of Philosophy (PhD)

<p>This thesis concerns itself with advances in Space Mapping optimization of microwave circuits and with the developments in Parameter Extraction. Space Mapping (SM) optimization aims at efficiently optimizing microwave circuits using the accurate and time-intensive electromagnetic simulators. Such simulators represent "fine" models of the circuit under consideration. SM exploits the existence of a less accurate but fast "coarse" model, e.g., an empirical model. A mapping is established between the parameter spaces of the coarse and fine models. The fine model design is the inverse mapping of the optimal coarse model design. A crucial step for any SM-based optimization algorithm is Parameter Extraction (PE). In this step a coarse model point that corresponds to a given fine model response is obtained through an optimization process. The nonuniqueness of PE can lead to divergence or oscillation of the optimization iterates. We introduced the Trust Region Aggressive Space Mapping (TRASM) algorithm. This algorithm integrates a trust region methodology with SM optimization. The iterate is confined to a trust region in which the utilized linearization can be trusted. TRASM also exploits a recursive multi-point parameter extraction step to enhance the uniqueness of PE. The Aggressive Parameter Extraction (APE) algorithm addresses the optimal selection of parameter perturbations used to increase trust in PE uniqueness. We establish an appropriate criterion for the generation of these perturbations. The APE algorithm classifies possible solutions for the PE problem. Two different approaches for obtaining subsequent perturbations are utilized based on a classification of the extracted parameters. The algorithm is demonstrated through parameter extraction of microwave filters and transformers. The Hybrid Aggressive Space Mapping (HASM) algorithm addresses the case of a poor coarse model. HASM utilizes SM optimization as long as it is converging. Otherwise, it switches to a direct optimization phase. We developed a relationship that relates the established mapping and the first order derivatives of the coarse and fine models. This relationship is utilized in switching between the SM phase and the direct optimization phase. We also present a Surrogate Model-based Space Mapping (SMSM) optimization algorithm. SMSM integrates two approaches for efficient optimization: SM optimization and surrogate model optimization. It exploits a surrogate model in predicting new iterates. This model is a convex combination between a mapped coarse model and a linearized fine model. The mapped coarse model exploits a frequency-sensitive mapping. During the optimization iterates, the coarse and fine models are simulated at two different sets of frequencies. Utilizing a frequency sensitive mapping is shown to enhance the uniqueness of PE. It also overcomes severe frequency misalignments between the responses of both models.</p> / Doctor of Philosophy (PhD)

<p>Adaptive filters are used in linear estimation problems when no a priori knowledge of signal statistics is available or when systems are time varying. They have become a popular signal processing tool and have found application in many diverse areas. The work in recent years has led many to regard the recursive least-squares algorithm (RLS) and its variants as the state of the art. The principal advantage of RLS over the rivalling least mean square algorithm (LMS) is its fast rate of convergence, which in many applications justifies the higher computational complexity of RLS. Mathematically speaking, both RLS and LMS result from the application of particular iterative optimization algorithms to the minimization of the least-squares criterion. The purpose of this thesis is to investigate the applicability of a new class of optimization methods, interior point optimization algorithms, to the adaptive filtering problem. We develop a new recursive algorithm, called Interior Point Least Squares or IPLS, that can be efficiently implemented with a computational cost comparable to (but higher than) RLS. IPLS matches RLS in asymptotic performance, but has a faster transient convergence rate. This is significant because until now "...[RLS'] convergence speed [has been] considered to be optimal in practice, and [thus] a measure for comparison for other algorithms.", (Moustakides, in a paper in the IEEE Transactions on Signal Processing , October 1997). Additional properties of IPLS are its insensitivity to variations in initialization, numerical stability in the presence of limited precision arithmetic, and versatility in that it readily allows the inclusion of additional constraints on the weight vector. Some of the above mentioned properties are further investigated and exploited in applications to adaptive equalization and adaptive beamforming. Numerical simulations are used throughout the thesis to illustrate, confirm, and extend our analytical results.</p> / Doctor of Philosophy (PhD)

<p>In the past, the discrepancy between predicted and measured core loss values has led to the use of empirical scaling factors to improve correlation. Accurate prediction of core loss is important in the design of electrical machinery. Machine customers often demand guaranteed efficiency values at the quotation stage. If the guaranteed value is missed on delivery, they may impose penalties of as much as $5000 for every additional kW of loss. This thesis addresses one possible source of this discrepancy, that is, the failure to account for rotational iron losses. A significant portion of an induction motor stator is exposed to flux that rotates in the plane of the machine laminations. Iron losses in rotating magnetic fields differ from those obtained under alternating flux conditions and are not accurately estimated from alternating loss measurements. In recent years, a great deal of research has occurred for the explicit purpose of developing a standardized test for measuring rotational iron losses. Work in this field has been motivated by the opinion that rotational iron loss data will be used by machine designers to refine the accuracy of core loss predictions in rotating machines. The hypothesis of this dissertation is that if iron losses due to rotational flux in the stator are calculated rigorously, some portion of the discrepancy between tested and calculated values of no-load iron loss will be accounted for. To test this hypothesis, the iron losses due to the fundamental frequency variation of the flux density have been calculated in several ways. The stator losses were calculated using alternating loss data available from standard tests and by rigorously accounting for the losses caused by rotational flux. While some differences have been noted, they are too small to account for any major portion of the discrepancy between tested and calculated values of no-load core loss. The results of our investigation show the hypothesis to be false. If a standardized test apparatus for making rotational iron loss measurements were realized in the near future, it would benefit producers of electrical sheet and provide useful information to machine designers. However, as this investigation will show, the refinement to core loss calculation methodologies will not likely improve the correlation of tested to calculated values of core loss.</p> / Doctor of Philosophy (PhD)

<p>Wireless communications technology is presently undergoing a tremendous expansion, which is brought on by the proliferation of many diverse and very compelling applications. These trends are continually pushing the demand for substantially increased information capacity, which can only be realized through the development of novel communication techniques. In this context, we may mention a ground-breaking wireless communication technique that offers a tremendous potential to increase the information capacity of the channel, namely, the multi-transmit and multi-receive (MTMR) antenna system, which is popularized as the Bell-Labs Layered Space-Time (BLAST) architecture. In particular, the Diagonal-BLAST (D-BLAST) and the Vertical-BLAST (V-BLAST), developed by Bell labs of Lucent Technologies, permit signal processing complexity to grow linearly, with the capacity increase being made possible through the use of a large number of transmit and receive antennas. However, from a practical perspective, D-BLAST is inefficient for short packet transmissions due to its boundary space-time wastage. Meanwhile, V-BLAST suffers from error propagation due to deep fades in the wireless channel. In this thesis, we propose Turbo-BLAST, a novel multi-transmit and multi-receive antenna system that can handle any configuration of transmit and receive antennas. It presents a framework of simple yet highly effective random space-time transmission and iterative joint-decoding receivers for BLAST architectures. Specifically, we show that the embodiment of turbo principles and the BLAST architecture provides a practical solution to the requirement of high data-rate transmission in a reliable manner for future wireless communication systems.</p> / Doctor of Philosophy (PhD)

<p>The problem of estimation of angle-of-arrival (AOA) in multipath environments is addressed in this thesis. In particular, two new estimation techniques are developed. The first technique is based on the maximum likelihood estimation (MLE). This algorithm is unique in that a highly deterministic multipath signal model is used when formulating the likelihood function, which is then maximised with respect to the AOA. The deterministic multipath signal model that has been developed to describe the physics underlying the propagation of signals from a signal radio source to a receiver is much more complete than the general AOA model commonly used in other maximum likelihood formulations. This model makes use of the geometrical information and a priori knowledge of a number of physical parameters. By using the deterministic multipath signal model with the MLE estimator, one is essentially making more information available to the estimation process. The net result is that the estimator's performance can be greatly enhanced. The Cramer-Rao bounds that apply specifically to this model have been derived to provide a performance measure for the mean-squared errors (MSE) in the estimated AOAs.</p> <p>Although the MLE method is optimum in a statistical sense, the computational load of the nonlinear optimisation procedure inherently required by the MLE method is too heavy for real-time processing. Accordingly, we propose a novel approach to the AOA estimation problem, which is based on the use of an associative memory. The functionality of an associative memory is identical to that of the inverse mapping network. This provides a more comprehensive explanation for the rationale of exploiting the inverse mapping concept in the AOA estimation problem. In particular, the AOA problem is considered as a mapping from the space of AOA to the space of the sensor output. A nonlinear associative memory is used to form the inverse mapping from the space of sensor output to the space of AOA and this memory is realised using the generalised radial basis function (RBF) neural network. In the actual implementation of the RBF network for AOA estimation, the second order statistics of the signals are used as the input vector of the network. The use of second order statistics eliminates the need to deal with the initial phase of the signal. Furthermore, it is suitable for the application to the minimum redundant array. The RHF network is much more efficient in terms of computation than the MLE algorithm. This makes the RBF network attractive for real-time implementation.</p> <p>Simulations are carried out to understand the efficiency of the RBF neural network approach. The learning and estimation performance is inversely proportional to the number of learning samples and the number of hidden units. At relatively low SNR, the estimation performance of the RBF network becomes insensitive to both the number of learning samples and the number of hidden units. The estimation performance of both the MLE technique and the RBF network is also evaluated as functions of the number of snapshots and SNR. The performance of the MLE algorithm is consistent with the Cramer-Rao bound. The MLE method is more efficient in terms of estimation than a RBF network, provided that the search resolution used in the MLE method is sufficiently high. For equivalent computational complexity, the RBF network gives much better performance than the MLE method. In terms of estimation, with the same computational complexity, the MSE produced by the RBF network is much less than that produced by the MLE method. In summary, for the same performance, the computational complexity required by MLE method is much higher than that required by the RBF network. It follows that the advantage to be gained by using a RBF network for AOA estimation is a considerable reduction in computational complexity. Finally, both the MLE technique and the RBF network are validated using real data, which were collected using a 32-element sampled aperture antenna. The results obtained using the RBF network are very similar to those obtained using the MLE method.</p> / Doctor of Philosophy (PhD)

<p>Surface acoustic wave (SAW) devices are an enabling technology for high-performance wireless communication systems. They are able to meet performance specifications that are beyond the scope of competing technologies, particularly in the front-end of the receivers, and establish the ultimate performance that is achievable. Current spectrum congestion is forcing a move to higher operating frequencies, and this is leading to a search for low-cost SAW devices that are able to operate at frequencies above 2 GHz. To this end, there has been considerable interest in devices that employ the high acoustic velocity of diamond. Diamond, however, is not piezoelectric, and it must be layered with other materials such as zinc oxide (ZnO) to permit the electrical generation and detection of acoustic waves. There is therefore a need for modeling tools that accurately predict the behaviour of multi-layered SAW substrates in the presence of surface transducers and reflectors. This Thesis presents a study of SAW propagation and generation under infinite periodic grating structures on multi-layered ZnO/Diamond substrates. The study is based on the space harmonic method (SHM) and predicts the SAW behaviour under both open and shorted surface electrodes. Dispersion diagrams are obtained around the first Bragg wavenumber and stopbands of finite bandwidth are observed. The method is then extended to the generation of SAWS by interdigital transducers. Admittance curves and static capacitances are calculated. The physical propagation behaviour of ZnO/Diamond multi-layered substrates is also investigated. The displacement distributions and the standing wave patterns are calculated within each layer. The energy contained in each layer is computed for different propagation modes and different ZnO layer thicknesses. The results are interpreted within the framework of the coupling-of-modes (COM) theory. The COM parameters are derived for the first and second Sezawa modes as a function of aluminum and zinc oxide thicknesses. The results and the COM parameters can be directly used in the design of SAW devices. The established analytical treatments can be easily applied to other multi-layered substrates with an arbitrary configuration including additional layers.</p> / Doctor of Philosophy (PhD)

<p>In a multi-user communication system such as the wireline or wireless communication systems, a commonly encountered problem is the extraction of the desired signal from Co-Channel Interference (CCI) and Adjacent Channel Interference (ACI). To combat the CCI and ACI, the conventional filtering techniques are unable to carry out the job. The optimum FREquency-SHift (FRESH) filtering technique proposed by W. A. Gardner enables us to suppress spectrally overlapped signals by using the cyclostationarity of the signals. However, to design the optimum FRESH filter, we must have the statistical knowledge of the desired signal or a training signal which, in practice, are not often available. This thesis proposes a blind adaptive FRESH filtering algorithm which does not need a training signal to extract the desired signal from spectrally overlapping interference. We call this new technique Blind Adaptive (BA)-FRESH filtering. Comparing the BA-FRESH filter with the FRESH filter with a training signal which is called Trained Adaptive FRESH (TA-FRESH) filter, it has been proved that BA-FRESH and TA-FRESH have same performances when the data length is infinite. On the other hand, various cyclic beamforming techniques such as the spectral Self-COherence REstoral (SCORE), the Cyclic Adaptive Beamforming (CAB), the Constrained Cyclic Adaptive Beamforming (C-CAB) and the Robust Cyclic Adaptive Beamforming (R-CAB) algorithms can be used to combat CCI and ACI efficiently. However, when the desired signal and the interferences are very closely spaced in arrival directions, system performance improvement using these cyclic beamforming alone is limited because the beamformers are just spatial filters. By combining the spatial beamforming with the temporal FRESH filtering, a large system performance improvement may be achieved due to the full utilization of the signal information in both time and space domains. A Blind Adaptive Space-Time (BLAST) algorithm is proposed in this thesis. The BLAST algorithm is a blind adaptive time varying space-time filter. The BLAST algorithm can be viewed as the expansion of the BA-FRESH filtering algorithm to the space-time domain. Comparing the BLAST filter with the space-time filter with a training signal which is called Trained Adaptive Space-Time (TAST) filter, it has been proved that BLAST and TAST have same performances when the data length is infinite. When the data length is finite, there are performance differences between BLAST and TAST. Convergence of the BLAST and TAST filter coefficients, the finite sample output signal to interference plus noise ratio (SINR), and the finite sample output mean square error (MSE) are analyzed. (Abstract shortened by UMI.)</p> / Doctor of Philosophy (PhD)