<p>This thesis addresses itself to computer oriented techniques for large scale optimization of analog circuits. New techniques for simulation and sensitivity analysis are described and are used to improve the performance of circuit optimization. A powerful automatic decomposition technique is developed directly enabling a normal optimizer to solve large circuit problems. Our theory is applied to the design of microwave circuits.</p> <p>The status of large scale circuit optimization and the state-of-the-art of microwave CAD are reviewed. The necessity of circuit oriented optimization techniques is demonstrated by formulating design, modelling, diagnosis and tuning into optimization problems.</p> <p>A comprehensive treatment of large change sensitivity computation for linearized circuits using generalized Householder formulus is presented. A technique for circuit response updating via a minimum order reduced system is developed. By avoiding re-analysis of the complete circuit, our method is responsible for efficient simulation of large circuits when a subset of the circuit parameters is frequently perturbed.</p> <p>An elegant theory for simulation and exact sensitivity analysis of branched cascaded networks is described. Our approach explicitly takes the circuit-structure into consideration and does not deteriorate as the overall network becomes large. The practicality of the theory is illustrated by efficient optimization of microwave multiplexers consisting of multi-cavity filters distributed along a waveguide manifold. Examples of optimizing 12- and 16-channel multiplexers are provided.</p> <p>A novel and general automatic decomposition technique for large scale optimization of microwave circuits is presented. The partitioning approach proposed by Kondoh for FET modelling problems is verified. The application of our technique is demonstrated by the large scale optimization of a 16-channel multiplexer involving 399 nonlinear functions and 240 variables.</p> / Doctor of Philosophy (PhD)

<p>The problem of self-tuning reference signal tracking is considered for systems represented by autoregressive moving average (ARMA) as well as state-space models. By self-tuning control, it is meant to be a combination of recursive parameter estimation and control algorithm. A new strategy of controller design is proposed, which is pole/zero placement in the 'error transfer function' (ETF) in contrast with the usual closed-loop pole-placement. Sufficient conditions for arbitrary simultaneous assignment of ETF poles and zeros are derived. For ARMA models, a recursive extended least squares type algorithm with a general nonlinear criterion function, which can be defined by the user, is suggested and the strong consistency of the algorithm is proved. Reference signal model identification is introduced for the first time into the context of adaptive control, which provides great flexibility to track any unknown external reference trajectory. The global convergence of the adaptive ETF pole/zero placement is theoretically established for deterministic systems. New stochastic optimal control algorithms are derived for the case where the control objective is reference signal tracking. The novelty of the proposed algorithms is that the performance indices are determined by the prespecified locations of ETF poles as well as zeros. State-space approach to self-tuning control has been studied also. The recursive prediction error method is used for joint state and parameter estimation, of state-space innovations model. Adaptive reference signal tracking control laws are derived for system output as well as an immeasurable physical state.</p> <p>To demonstrate practical applications, the derived self-tuning algorithms were applied to surface accuracy control in turning and end milling process. The results of simulations indicate considerable improvements in geometric accuracy of finished workpieces over conventional numerical control in the presence of significant tool workpiece deflection.</p> / Doctor of Philosophy (PhD)

<p>The major purpose of the work described in this thesis is to characterize the basic properties and evaluate the potential of optical fibre filters. As it turns out, these filters have a spectral bandwidth in reflection which can be varied over a very wide range, from as narrow as 0.01 to several angstroms (A). As a result, they are potentially very important for applications as multiplexers and demultiplexers in Wavelength-Division-Multiplexing (WDM) in Optical Communication Systems (OCS) (WDM is a means of increasing the data handling capability of a given fibre system through the use of multiple propagating wavelengths). Up to the present moment, gratings in the form of slabs have invariably been employed as the multiplexers and demultiplexers in WDM which greatly restricts the number of wavelengths which can be used since a large spectral bandwidth separation (>100 Ǻ) is required between adjacent channels is such systems.</p> <p>The optical fibre filters we report here can have a spectral bandwidth narrower than 1 Ǻ and consequently, employing them as multiplexers and demultiplexers in WDM allows a lot more channels to be transmitted. Other than this particular application, optical fibre filters can equally well function as resonant reflectors in a laser cavity, distributed feedback devices. tunable filters or external wavelength selective reflectors for heterojunction injection laser by end butting them to the latter. Furthermore, they have the potential to be used for equalizing material dispersion in OCS employing single mode fibres. Lastly, they can be tailored to give multifilters, comb filters or in general, filters that perform a variety of specified functions.</p> <p>The approach to the characterization of optical fibre filters is divided into two parts. The first part deals with the development of a theoretical model that can be used to predict and describe the performance of the optical fibre filters. The second part describes the experiments for the actual fabrication of the filters and subsequent extraction of some of their physical parameters. By comparing the theory with the experimental results, we are able to deduce the range of filters available when using an Argon Iaser as the means of writing filters and subsequently evaluate their full potential in OCS. Also, a very important relationship between the optical writing power applied to a fibre and the resultant index perturbation in the fibre or equivalently, the resultant filter characteristics, is established so that the actual fabrication of filers designed to satisfy specific function can be performed.</p> / Master of Engineering (ME)

<p>With the availability of inexpensive computer hardware, software intensive systems are becoming sophisticated and pervasive, creating a need for software design methods that deliver robust and efficient systems. Unfortunately, most systems today are designed by trial and error, with designers having little insight into their correctness before they are implemented. The hallmark of an engineering method is the use of system models (prototypes) to verify designs by provably establishing essential characteristics of products before they are constructed. This thesis attempts to bring the craft of software construction closer to an engineering discipline, by transforming theoretical ideas in program verification and concurrency theory into a concrete method for the analysis of software requirements and system specifications. In our opinion, such methods will find wider acceptance if they are adequately supported by a set of tools. This thesis also describes tools that are being developed to support our verification method.</p> <p>In this thesis, our primary concern is the problem of establishing that systems, when constructed in accordance with a design, will meet their required correctness properties. We describe a verification method, TOP, in which the description of a system (i.e., its design) is expressed as an abstract program in a programming language-like notation called MELA. We represent essential characteristics of the system (or its requirements) as predicates in a formal logic. The problem of establishing required correctness properties for the system is then reduced to the problem of establishing that a set of logical formulae hold. This is known as the verification problem. Our verification method, TOP, is mainly applicable to systems with complex control structures and simple data structures, and is tailored to systems that are designed not to terminate, and which often exhibit nondeterministic behaviour i.e., different runs of the system may produce different results, for the same inputs. Examples of such systems are operating systems, computer hardware, communications protocols, telecommunications systems, and control systems.</p> <p>Our method offers a unified framework in which correctness properties may be verified using a combination of model checking (a fully automatic method), and theorem proving (a partially automated approach). This is desirable, because automated methods are applicable only to a limited class of systems. They are severely limited by the size of the state space generated by the abstract program denoting the system description - current limits are in the region of 108 states - roughly the number of states that could be generated by a program with a single 32-bit integer variable. Theorem proving approaches, while more general, are tedious and require specialised knowledge (the ability to provide proofs) on the part of verifiers. When carried out manually, proofs also tend to be error-prone. Theorem proving is particularly valuable in cases where the effort invested in carrying out a verification may be amortised over several projects (good examples are communications protocol specifications and distributed algorithms). Theorem proving can benefit tremendously from appropriate mechanical support, which can automate the tedious parts, in addition to playing the part of a relentless skeptic who demands the utmost precision and rigour in proofs. By presenting verifiers with a unified framework, therefore, we hope to extend the range of verification to systems that cannot be verified by either method alone.</p> <p>One of the common complaints about theorem proving systems is that they are bewildering to beginners and hard to use. In this thesis we present an improved user interface for theorem provers. We also describe the details of system SNAP which has been built with this interface. SNAP has been designed for the purpose of checking proofs, in addition to providing partial automation of proofs. SNAP additionally allows users to carry out proofs at the desired level of abstraction. To support this, SNAP includes proof libraries and simplification algorithms based on conditional term rewriting.</p> <p>Our method, TOP, has been used to verify problems derived from practice. We present two case studies that involved the use of TOP. In the first study, we provide a formal semantics for a tabular notation for requirements documents in terms of MELA. We then use model checking and theorem proving to establish certain "safety assertions" for the requirements specification of a system that mediates access to data shared by two processes. In the second study, we analyse liveness violations in a communications protocol standard, and verify that suggested changes to the standard have indeed fixed the problem. Finally, we conclude by describing ongoing work and suggestions for future research.</p> / Doctor of Philosophy (PhD)

<p>The value of the time domain simulation technique for power system transient stability analysis can be greatly enhanced if it can indicate the degree of stability by producing a stability index. Often referred to as Energy Margin, the stability index offers additional insight into the transient stability problem being studied, and has the capability of speeding up transient stability limit derivations. Moreover, energy margins also have potential applications in dynamic contingency ranking and screening.</p> <p>This thesis presents a new method, called the Relevant Fault-on Trajectory (RFT) method, for incorporating energy margin calculations into time domain simulations. The proposed method is based on the determination of the additional energy absorbing capability of the critical generator group, at the instant when the transient kinetic energy injected into this group by the disturbance is fully absorbed. The additional energy absorbing capability is obtained through the simulation of a relevant fault-on trajectory. The RFT method computes energy margins efficiently and reliably for systems exhibiting either plant mode or area mode stability problems.</p> <p>The practicality of the proposed method has been successfully demonstrated on a 27-generator, an 89-generator, and a 144-generator system. The RFT method has the capability of speeding up transient stability limit derivation by reducing the number of stability runs.</p> / Doctor of Philosophy (PhD)

<p>A feasibility study has been performed on a 4 bits/s/Hz digital radio system. This system utilizes a 4-level quadrature amplitude modulation scheme with Duobinary filtering to achieve such high bandwidth efficiency. The error rate performance of such systems is evaluated. A repeater structure including the transmitter and the receiver of such systems is developed and analyzed. A decision directed carrier recovery loop is presented and analyzed using stochastic techniques. Expressions for the effective phase detector characteristic and the steady state phase error probability density function are derived. A symbol timing recovery structure is also presented and exhaustive simulations have been performed to analyze the performance of the timing recovery structure.</p> / Master of Engineering (ME)

<p>Surface acoustic wave (SAW) resonators, longitudinally coupled resonators and waveguide coupled resonators are studied in this thesis. The operation principles of these devices are examined. Standing wave pattern analyses are used in explaining the operation of the longitudinally coupled resonators. The coupling-of-modes (COM) theory is extensively used to characterize these devices. The COM theory is extended to the analysis of the coupled gratings and coupled transducers in which multi-track coupling is involved. The reflection, transduction and attenuation for the gratings and transducers are described in the COM model for these elements. Closed-form solutions of the COM equations for uniform structures are obtained. The matrix expression for these solutions, because of its modular nature, makes it very easy to cascade the different elements in a device and to analyze devices with different structures. The admittance matrix and scattering matrix can be calculated over the frequency range of interest. Efficient and accurate models are developed for these devices. There is good agreement between the simulated results and the experimental ones. The design procedure and optimization of waveguide coupled resonators are discussed and demonstrated. The longitudinal field distribution of the waveguide coupled resonator is calculated at the resonant frequencies of the devices. It is demonstrated that the extended COM theory can be used in the analysis of waveguide coupled resonators with more than two tracks. A four-track waveguide coupled resonator is analyzed using the extended COM theory. Based on the understanding of the waveguide and longitudinally coupled resonators, a four-pole resonator with both longitudinal coupling and waveguide coupling is proposed and analyzed.</p> / Doctor of Philosophy (PhD)

<p>Time Division Multiple Access (TDMA) or burst mode operation of a communication system requires that synchronization be achieved within an alloted preamble time or else the entire data burst may be lost. A conventional phase-locked loop occasionally exhibits a prolonged lock-up time known as hangup, especially when the initial phase error falls within the neighbourhood of 180 degrees, and is therefore not suitable for synchronization of this type of system. The main purpose of this work is to extend the investigation of a modified phase-locked loop which is termed a limit-switched loop (LSL). The LSL offers a faster lock-up time and lower probability of hangup. The transient probability density function (pdf) of the phase error for a first-order LSL and the steady-state pdf of the phase error for the second-order LSL have been obtained, using Fokker-Planck techniques. Simulations and experimental methods have been used to obtain the probability of hangup on a low frequency model which also reveals the optimum range for the switching time.</p> / Master of Engineering (ME)

<p>A digital signal processor, operating at baseband, is described for the scanning of a Mills' cross array antenna. The processor uses time modulation to scan the pencil beam produced by the array in a raster-like fashion over the prescribed sector. The scheme uses within-pulse scanning in order to avoid the loss of power in the fan beams outside the area covered by the pencil beam. This processor provides flexibility in applying different types of aperture distribution to modify the side-lobe structure of the antenna pattern. It provides a simple means for changing the scanning rate of the array.</p> <p>The different types of scanning technique are reviewed briefly. A detailed mathematical analysis of time modulation scanning technique is presented. Experimental results are included to demonstrate the practical feasibility of the digital baseband processor for scanning the Mills' cross array antenna along prescribed look directions inside the visible region of interest.</p> / Master of Engineering (ME)

<p>An interface between a NOVA 830 and IMSAI 8080 microcomputer is presented. The design and implementation of this interface will enable the asynchrous interchange of information between the mini and minicomputer mentioned.</p> <p>The design presented in this thesis will enable the bidirectional transfer of data blocks via DMA. Besides the capability to generate DMA cycles on both computer buses, a portion of the NOVA 830 memory can be shared. Under this shared memory arrangement, the IMSAI 8080 microcomputer can access the shared NOVA 830 memory as though it is directly connected to its system bus.</p> <p>Along with the circuit design of the interface, the software requirements involved to ultilize the interface are also discussed.</p> / Master of Engineering (ME)