by Cheng Sze Wan. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1999. / Includes bibliographical references (leaves 72-[76]). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Background --- p.1 / Chapter 1.2 --- Traffic Scheduling in Input-buffered Switches .。 --- p.3 / Chapter 1.3 --- Organization of Thesis --- p.7 / Chapter 2 --- Principle of Enchanced PIM Algorithm --- p.8 / Chapter 2.1 --- Introduction --- p.8 / Chapter 2.1.1 --- Switch Model --- p.9 / Chapter 2.2 --- Enhanced Parallel Iterative Matching Algorithm (EPIM) --- p.10 / Chapter 2.2.1 --- Motivation --- p.10 / Chapter 2.2.2 --- Algorithm --- p.12 / Chapter 2.3 --- Performance Evaluation --- p.16 / Chapter 2.3.1 --- Simulation --- p.16 / Chapter 2.3.2 --- Delay Analysis --- p.18 / Chapter 3 --- Providing Bandwidth Guarantee in Input-Buffered Switches --- p.25 / Chapter 3.1 --- Introduction --- p.25 / Chapter 3.2 --- Bandwidth Reservation in Static Scheduling Algorithm --- p.26 / Chapter 3.3 --- Incorporation of Dynamic and Static Scheduling Algorithms .。 --- p.32 / Chapter 3.4 --- Simulation --- p.34 / Chapter 3.4.1 --- Switch Model --- p.35 / Chapter 3.4.2 --- Simulation Results --- p.36 / Chapter 3.5 --- Comparison with Existing Schemes --- p.42 / Chapter 3.5.1 --- Statistical Matching --- p.42 / Chapter 3.5.2 --- Weighted Probabilistic Iterative Matching --- p.45 / Chapter 4 --- EPIM and Cross-Path Switch --- p.50 / Chapter 4.1 --- Introduction --- p.50 / Chapter 4.2 --- Concept of Cross-Path Switching --- p.51 / Chapter 4.2.1 --- Principle --- p.51 / Chapter 4.2.2 --- Supporting Performance Guarantee in Cross-Path Switch --- p.52 / Chapter 4.3 --- Implication of EPIM on Cross-Path switch --- p.55 / Chapter 4.3.1 --- Problem Re-definition --- p.55 / Chapter 4.3.2 --- Scheduling in Input Modules with EPIM --- p.58 / Chapter 4.4 --- Simulation --- p.63 / Chapter 5 --- Conclusion --- p.70 / Bibliography --- p.72

by Vincent Wing-Shing Tse. / Thesis submitted in: December 1997. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1998. / Includes bibliographical references (leaves 62-[64]). / Abstract also in Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Principles of Multirate Circuit Switching in Quantized Clos Network --- p.10 / Chapter 2.1 --- Formulation of Multirate Circuit Switching --- p.11 / Chapter 2.2 --- Call Level Routing in Quantized Clos Network --- p.12 / Chapter 2.3 --- Cell Level Routing in Quantized Clos Network --- p.16 / Chapter 2.3.1 --- Traffic Behavior in ATM Network --- p.17 / Chapter 2.3.2 --- Time Division Multiplexing in Multirate Circuit Switching and Cell-level Switching in ATM Network --- p.19 / Chapter 2.3.3 --- Cell Transmission Scheduling --- p.20 / Chapter 2.3.4 --- Capacity Allocation and Route Assignment at Cell-level --- p.29 / Chapter 3 --- Performance Evaluation of Different Implementation Schemes --- p.31 / Chapter 3.1 --- Global Control and Distributed Switching --- p.32 / Chapter 3.2 --- Implementation Schemes of Quantized Clos Network --- p.33 / Chapter 3.2.1 --- Classification of Switch Modules --- p.33 / Chapter 3.2.2 --- Bufferless Switch Modules Construction Scheme --- p.38 / Chapter 3.2.3 --- Buffered Switch Modules Construction Scheme --- p.42 / Chapter 3.3 --- Complexity Comparison --- p.44 / Chapter 3.4 --- Delay Performance of The Two Implementation Schemes --- p.47 / Chapter 3.4.1 --- Assumption --- p.47 / Chapter 3.4.2 --- Simulation Result --- p.50 / Chapter 4 --- Conclusions --- p.59 / Bibliography --- p.62

by Wai-Shan Chan. / Thesis submitted in: August 1997. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1998. / Includes bibliographical references (leaves 61-63). / Chapter Chapter 1 --- Introduction / Chapter 1.1 --- Motivation --- p.1 / Chapter 1.2 --- Fiber recirculating delay loop --- p.2 / Chapter 1.3 --- Slicing scheme --- p.3 / Chapter 1.4 --- Outline of the thesis --- p.4 / Chapter Chapter 2 --- Fiber recirculating delay loop / Chapter 2.1 --- Review --- p.5 / Chapter 2.1.1 --- Introduction --- p.5 / Chapter 2.1.2 --- The device --- p.6 / Chapter 2.1.3 --- Filtering properties of the fiber recirculating delay loop --- p.8 / Chapter 2.1.4 --- Noise properties of the fiber recirculating delay loop --- p.10 / Chapter 2.1.5 --- Limitations of the BRL device --- p.13 / Chapter 2.2 --- Discussion --- p.14 / Chapter 2.3 --- Summary --- p.14 / Chapter Chapter 3 --- Slicing scheme / Chapter 3.1 --- Introduction --- p.16 / Chapter 3.2 --- Slicing Scheme --- p.18 / Chapter 3.3 --- Experimental Investigation --- p.20 / Chapter 3.4 --- Numerical analysis --- p.33 / Chapter 3.5 --- Simulations --- p.40 / Chapter 3.6 --- Discussion --- p.49 / Chapter 3.6.1 --- The relationship between the system penalty and the ratio fd/fs --- p.49 / Chapter 3.6.2 --- Design of BRL --- p.51 / Chapter 3.6.3 --- Advantages and Disadvantages of the slicing scheme --- p.56 / Chapter 3.7 --- Summary --- p.56 / Chapter Chapter 4 --- Conclusion / Chapter 4.1 --- Fiber recirculating delay loop as a BRL device --- p.58 / Chapter 4.2 --- Slicing scheme --- p.59 / Chapter 4.3 --- Future work --- p.60 / Bibliography --- p.61

The main inquiry for this research is to identify the reasons that contribute to customer switching intention decisions. In addressing this enquiry, two theories - the theory of migration and theory of planned behaviour - were identified as the theoretical framework underpinning the study. Two main objectives of the study were (i) to identify the push-pull and mooring factors and (ii) to measure the relationship between the push-pull and mooring factors towards switching intention. The investigation focused on the impact of with push-pull-mooring factors on switching intention. Early research into switching behaviour studies focused largely on variables that contribute to the switching intention decision, mainly due to the critical incidents encountered by customers that push them from their origin or pull them to another destination or mooring factors that might inhibit or mitigate their switching decision. In view of this, a combination of push-pull and mooring variables were used to measure the switching intention behaviour. A multiple method approach was used to study the issues in two different stages. In the first stage qualitative data collection was used to support and confirm the identification of factors from the literature. For the main quantitative methods, using a hypothetical deductive testing approach, this study (N=2018) used survey data collected via a self-administered, voluntary online survey, to develop switching intention behaviour model. The results indicated that situational factors, positive attachments, perceived switching benefits, positive attitudes towards switching and positive beliefs of others towards switching emerged as consistent push factor while availability of alternatives emerged as the pull factors. Interestingly poor pricing, poor service incidents, positive ability to switch and switching barriers were not supported in this study indicating that there is no relationship between poor pricing, poor service incidents, positive ability to switch and switching barriers towards switching intention.

Zheng, Pengxuan. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2007. / Includes bibliographical references (leaves 53-57). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgments --- p.v / Table of Contents --- p.vi / List of Figures --- p.viii / List of Tables --- p.ix / Chapter Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Motivation --- p.1 / Chapter 1.2 --- Related Work --- p.2 / Chapter 1.3 --- Our Contribution --- p.3 / Chapter 1.4 --- Organization of the Thesis --- p.4 / Chapter Chapter 2 --- Background Overview --- p.6 / Chapter 2.1.1 --- Traditional Wireless Networks --- p.6 / Chapter 2.2 --- Exponential Backoff --- p.7 / Chapter 2.2.1 --- Introduction --- p.7 / Chapter 2.2.2 --- Algorithm --- p.8 / Chapter 2.2.3 --- Assumptions --- p.9 / Chapter 2.3 --- System Description --- p.9 / Chapter 2.3.1 --- MPR Capability --- p.9 / Chapter 2.3.2 --- Backoff Slot --- p.10 / Chapter 2.3.3 --- Carrier-sensing and Non-carrier-sensing Systems --- p.11 / Chapter Chapter 3 --- Multipacket Reception in WLAN --- p.12 / Chapter 3.1 --- MAC Protocol Description --- p.13 / Chapter 3.2 --- Physical Layer Methodology --- p.16 / Chapter 3.2.1 --- Blind RTS Separation --- p.17 / Chapter 3.2.2 --- Data Packet Detection --- p.19 / Chapter Chapter 4 --- Exponential Backoff with MPR --- p.21 / Chapter 4.1 --- Analytical Model --- p.22 / Chapter 4.1.1 --- Markov Model --- p.22 / Chapter 4.1.2 --- Relations betweenpt andpc --- p.23 / Chapter 4.2 --- Simulation Settings --- p.26 / Chapter 4.3 --- Asymptotic Behavior of Exponential Backoff --- p.27 / Chapter 4.3.1 --- Convergence ofpt andpc --- p.27 / Chapter 4.3.2 --- Convergence of Npt --- p.29 / Chapter Chapter 5 --- Non-carrier-sensing System --- p.31 / Chapter 5.1 --- Performance Analysis --- p.31 / Chapter 5.1.1 --- Throughput Derivation --- p.31 / Chapter 5.1.2 --- Throughput Analysis --- p.32 / Chapter 5.1.3 --- Convergence of S --- p.36 / Chapter 5.2 --- Infinite Population Model --- p.38 / Chapter 5.2.1 --- Attempt Rate --- p.38 / Chapter 5.2.2 --- Asymptotic Throughput of Non-carrier-sensing System --- p.39 / Chapter Chapter 6 --- Carrier-sensing System --- p.43 / Chapter 6.1 --- Throughput Derivation --- p.43 / Chapter 6.2 --- Asymptotic Behavior --- p.44 / Chapter Chapter 7 --- General MPR Model --- p.48 / Chapter Chapter 8 --- Conclusions --- p.51 / Bibliography --- p.53

The vibration control problem for flexible structures is examined within the context of overall controller performance and power reduction. First, the issue of optimal sensor and actuator placement is considered along with its associated control robustness aspects. Then the option of alternately activating subsets of the available devices is investigated. Such option is considered in order to better address the effects of spatiotemporally varying disturbances acting on a flexible structure while reducing the overall energy consumption. Towards the solution to the problem of optimal device placement, three different approaches are proposed. First, a computationally efficient scheme for the simultaneous placement of multiple devices is presented. The second approach proposes a strategy for the optimal placement of sensors and collocated sensor/actuator pairs, taking into account the influence of the spatial distribution of disturbances. The third approach provides a solution to the actuator location problem by incorporating considerations with respect to preferred spatial regions within the flexible structure. Then the second problem named above is considered. Activating a subset of the available and optimally placed actuators and sensors in a flexible structure provides enhanced performance with reduced energy consumption. Such approach of switching on and off different actuating devices, depending on their local-in-time authority, results in a hybrid system. Therefore the proposed work draws on existing results on hybrid systems and includes an additional degree of freedom, whereby both the actuating devices and the control signals allocated to them are switched in and out. To enable this switching an activation strategy, which insures also that stability-under-switching is guaranteed, is required. Three different strategies are considered for such actuators allocation: first a cost-to-go index is considered, then a cost function based on the mechanical energy of the flexible structure and finally a performance index based on the maximum deviation of the transverse displacement. A flexible aluminum plate was chosen to validate and test the proposed approaches. The set up utilized four pairs of collocated piezoceramic patches that serve to provide sensing and actuating capabilities. Extensive numerical simulations were performed for both the placement strategies and the switching policies proposed, in order to predict the behavior of the flexible plate and provide the optimal actuator and sensor locations that were to be affixed on the flexible structure. Finally, to complete the validation process a sequence of experimental tests were performed. The objective of these tests was to compare the performance of the proposed hybrid control system to traditional non switched control schemes. In order to provide a repeatable perturbation, four of the piezoceramic patches were allocated to simulate a spatiotemporally varying disturbance, while the remaining four patches were used as sensors and controlling actuators. The experimental results showed a significant performance improvement for the switched controller over the traditional controller. Moreover the switched controller exhibited improved robustness towards spatiotemporally varying disturbances while the traditional controller showed a significant loss of controller performance. The improvement achieved in vibration control problems could be extended to a wider range of applications. In particular, although this study was concentrated on a rectangular thin plate, the proposed strategies can be applied to emph{any} structure and more generally to any plant whose dynamics can be represented by a second order linear system. For example, by removing the restriction of spatially fixed actuators and sensors, the proposed theory can be applied to the problem of unmanned vehicles control.

Gallium Nitride (GaN) power devices have the potential to jump-start the next generation of power converters which are smaller, faster, denser, and cheaper. They are thus expected to meet the increasing 21st Century need for power density and efficiency, while at the same time reducing pollution. With the commercialisation of 600 V GaN power devices, which the industry is keen to adopt, come significant challenges. Since there are a number of such devices which are new to the power community, there is a steep learning curve involved, with dispersed information on how best to employ these devices. This work aims to solve this problem through the development of a universal GaN power device and circuit model and the formulation of design rules and guidelines. Through this contribution, designers will be able to better understand and work with these novel devices with relative ease. This will aid the need for faster adoption of GaN devices by the industry solving the barriers to commercialisation. This research demonstrates the use of machine learning (ML) algorithms for behavioural modelling of GaN power devices. Introducing ML as the key to developing a general behavioural and circuit model for GaN power devices combined with understanding, learning, customizing and successfully demonstrating it is the major contribution of this research work. This research first presents a comprehensive investigation into the parasitic effect on the GaN device switching performance. A simple process based on RF techniques is introduced to approximately extract the impedances of the GaN device to develop a behavioural model. The switching behaviour of the model is validated using simulation and double pulse test experiments at 450 V, 10 A test conditions. The developed behavioural model for Transhporm GaN HEMT is 95.2% accurate as the existing LT-spice manufacturer model, and is very much easier for power designers to handle, without the need for knowledge about the physics or geometry of the device. However, given that separate models would need to be developed for each commercial GaN device, the need for a generalized and accurate GaN behavioural model was identified, and it is this generalised model that the remainder of this thesis focuses on. In the next part of this research, a GaN platform test bench is built through bridging RF and power electronics design methodologies to achieve a gate loop and power loop inductance of around 1.8nH with switching waveforms with rise time and fall time around 2.5ns at 450V, 15A, 500KHz test conditions. The double pulse test circuits are customized using different off the shelf gate drives and analysed for collecting switching data for training the ML model. ML modelling using supervised learning is used to predict the switching voltage and current waveforms thus making it possible to construct a generic GaN black box model. Different architectures with single and multi- layer neural networks are explored for modelling. The ability to demonstrate a GaN device ML model that maps both voltage and current inputs and outputs is another characteristic and novel feature of this work. This research demonstrates different types of GaN ML models. The developed voltage and current prediction models are based on feed forward neural network (FFNN), long short-term memory unit (LSTM) and gated recurrent unit (GRU). Several parameters are quantified and compared for validating the models. They are the network architectures, parameters, training time, validation loss and error loss. The ML models are also compared with the demonstrated model of chapter 3 and existing LT-Spice manufacturer models. The results show that the author has been able to develop a GaN LSTM ML model with an error rate of 0.03, and convergence at 3s with excellent stability. The ML based modelling is then translated from GaN power devices to GaN based circuits. Among the different neural network architectures trained and tested, a multi FFNN with 5 hidden layers and 30 neurons, was found to be the best for prediction and optimization. The switching behaviour comparison results shows the benefits and value of ML modelling in opening up whole new possibilities of employing advanced control algorithms for very efficient, reliable and scalable performance of GaN power electronics systems. Finally, the findings of this work have been generalized to frame machine learning based techniques to address the need for generic modelling of power electronic devices. These solutions are presented as an information manual to researchers, engineers and students interested in benefiting from adopting machine learning for power electronics applications.

É apresentado um amplificador de potência (PA) com controle digital da potência de saída, operando na banda S de frequência (2,2 GHz). Este PA utiliza um transformador de entrada para reduzir as flutuações dos sinais de terra. Um estágio de excitação oferece uma impedância apropriada para a fonte de entrada e ganho para o próximo estágio. O estágio de controle é usado para melhorar a eficiência do PA, composto por quatro ramos paralelos de chaves, onde os estados (ligado ou desligado) são separadamente ativados por uma palavra de controle de 4 bits. O estágio de saída implementa um amplificador classe E, usando uma topologia cascode para minimizar o estresse de tensão sobre os transistores, permitindo sua utilização sob tensão de alimentação de 3,3 V para se atingir uma potência de saída máxima em torno de 1 W, em um processo CMOS 130 nm, cuja tensão típica de alimentação é 1,2 V. O PA proposto foi projetado em uma tecnologia CMOS 130 nm para RF, ocupa uma área de 1,900 x 0,875 mm2 e os resultados das simulações em leiaute extraído obtidos demonstram uma potência de saída máxima de 28,5 dBm (707 mW), com PAE (Power- Added Efficiency) correspondente de 49,7%, para uma tensão de alimentação de 3,3 V. O controle de 4 bits permite um ajuste dentro da faixa dinâmica da potência de saída entre 13,6 a 28,5 dBm (22,9 a 707 mW), divididos em 15 passos, com o PAE variando de 9,1% a 49,7%. O PA proposto permite redução do consumo de potência quando este não está transmitindo na potência máxima. A potência consumida atinge um mínimo de 0,21Wquando a potência de saída é de 13,6 dBm (22,9 mW) e um máximo de 1,4 W quando a potência de saída é de 28,5 dBm (707 mW), o que representa 1,19 W de economia, aumentando a vida da bateria. A linearidade obtida neste circuito mostrou-se suficiente para atender os requisitos da máscara de emissão de espúrios de um padrão de comunicação com envoltória constante largamente utilizado, apresentando desempenho adequado para atender as especificações dos sistemas de comunicações modernos. / A power amplifier with digital power control useful to S-Band (2.2 GHz) applications and with an output power around 1 W is presented. It uses an input transformer to reduce ground bounce effects. A tuned driver stage provides impedance matching to the input signal source and proper gain to the next stage. A control stage is used for efficiency improvement, composed by four parallel branches where the state (on or off) is separately activated by a 4-bit input. The class-E power stage uses a cascode topology to minimize the voltage stress over the power transistors, allowing higher supply voltages. The PA was designed in a 130 nm RF CMOS process and the layout has a total area of 1.900 x 0.875 mm2, post-layout simulations resulted a peak output power of 28.5 dBm with a maximum power added efficiency (PAE) around 49.7% under 3.3 V of supply voltage. The 4-bit control allows a total output power dynamic range adjustment of 14.9 dB, divided in 15 steps, with the PAE changing from 9.1% to 49.7%. The proposed PA allows reduce the power consumption when it isn’t transmitting at the maximum output power. Where the power consumption is only 0.21 W when the PA is at the minimum output power level of 13.6 dBm (22.9 mW), which is 1.19 W smaller than the power consumption at full mode (1.4 W), increasing the battery life. The linearity in this circuit meet the emission mask requirements for a widely used communication standard with constant envelope. Post-layout simulation results indicate an overall performance adequate to fulfill the specifications of modern wireless communication systems.

Microfiche copy available in the Institute Archives and Barker Engineering Library. / Vita. / Thesis (Ph. D.)--Massachusetts Institute of Technology. Dept. of Engineering and Computer Science, 1977. / Includes bibliographical references. / by John Ippocratis Capetanakis. / Ph.D.

A method is proposed to infer stability properties for non-linear switching under continuous state feedback. Continuous-time systems which are dissipative in the multiple storage function sense are considered. A partition of the state space, induced by the cross-supply rates and the feedback function, is used to derive a restriction on switching. Then, conditions are proposed, under which, systems controlled by the feedback function and switching according to the rule are stable. In particular, Lyapunov and asymptotic stability are proved, both in a local and in a global context. Further, it is shown that the approach can be extended when one uses multiple controllers, and, therefore, is able to construct multiple partitions; conditions for this case are also presented. Finally, it is shown that, for the switching families that satisfy the switching rule posited by the results, one is able to find elements (that is, stabilising switching laws for the system) which are non-Zeno. Additional rule-sets that allow this are provided. It is argued that the conditions proposed here are easier to verify and apply, and that they offer additional flexibility when compared to those proposed by other approaches in the literature. The same infrastructure is used in the study of hybrid systems. For a general class of non-linear hybrid systems, a new property is proposed, that retains some of the properties of dissipativity, but it differs from it, crucially in the fact that it is not purely input-output. For systems having this property, it is shown that the partition used in the switching case can also be used. This, along with a set of conditions allows for the characterisation of the system behaviour in two scenaria. First, when the continuous behaviours and the jumping scheme act co-operatively, leading the system to lower energy levels (from the dissipativity point of view). Second, when the continuous behaviours are allowed to increase the stored energy, but the jumping is able to 6 compensate this increase. In the first case, it is shown that the equilibrium point under study is stable; in the second, it is shown that the system exhibits a type of attractivity, and, under additional conditions, it is asymptotically stable. Besides stability, a collection of stabilisation results are given for the case of dissipative switching systems. It is shown that one may design state feedback functions (controllers) with the objective that they satisfy the conditions of the stability theorems in this work. Then, systems under the designed controllers are shown to be stable, provided that the switching adheres to a specific switching rule. This problem is approached using a variety of tools taken from analysis, multi-valued functions and the space of non-switching stabilisation. In addition to the main results, an extensive overview of the literature in the area of switching and hybrid systems is offered, with emphasis on the topics of stability and dissipativity. Finally, a collection of numerical examples are given, validating the results presented here.