A problem in modern wireless communications is the scarcity of electromagnetic radio spectrum. The traditional fixed spectrum assignment strategy results in spectrum crowding on most frequency bands. Due to limited availability of radio spectrum and high inefficiency in its usage, cognitive radio networks have been seen as a promising solution to reducing current spectrum under-utilization while accommodating for the increasing amount of services demands and applications in wireless networks. Compared with the traditional networks, cognitive radio networks exhibit some distinct features, which result in necessity of further research in the resource allocation and scheduling that have been solved for the traditional networks. In this thesis, we focus on the packet scheduling in a single cell cognitive radio system with a single channel. An adaptive weight factor is introduced to adjust the priority of different cognitive radio users to be selected for service. The purpose of this research is to solve the unfairness problem of the traditional proportional scheduling schemes when used directly in a cognitive radio network, which lead to a user starved for a long time if it experiences a poor channel condition when the channel is available and experiences a good channel condition when the channel is not available. An adaptive weighted scheduling scheme is proposed to improve the performance in terms of throughput and fairness by jointly considering the instantaneous propagation conditions, adaptive weighted factor and the channel availability. The saturated traffic and non saturated traffic cases are considered. Some important performance metrics are investigated in the simulation, such as the system throughput, fairness, and service probability, and are quantified by the impact of weights and channel conditions. Extensive simulations have been conducted to demonstrate the effectiveness and efficiency of the proposed scheduling scheme.

Techniques and technologies to transfer electromagnetic energy through sub-wavelength channels have been researched extensively in the past few years because their application in different areas such as sub-wavelength imaging, telecommunication, increasing the storage capacity, and confinement and transmission of electromagnetic energy. Common ways of achieving such transmission includes exciting surface plasmon polaritons on both sides of the cannel or using double negative metamaterials. Recently a mechanism to squeeze the electromagnetic energy through sub-wavelength channels using materials with extremely small permittivity was introduced. Such materials may be found naturally at some limited frequencies in the infrared and optical frequency ranges, but they are commonly fabricated for a desired frequency as engineered metamaterials by by embedding metallic inclusions in a dielectric medium. The main problem with the engineered materials is that they have relatively large losses at their low permittivity frequency. In this thesis,I have presented a novel structure consisting of arrays of metallic wires that can be used to squeeze electromagnetic energy through sub-wavelength channels and junctions with negligible loss. The theory of transmission through such array is derived and design methods to tune the transmission frequency is provided. The structure is also tested numerically and experimentally in several geometries and results are compared with previous methods.

This thesis primarily covers the automation problem for sleep versus awake detection, which is sometimes accomplished by differentiating the various sleep stages prior to clustering. This thesis documents various experimentation into areas where the performance can be improved, including classifer design and feature selection from EEG, EOG and Context. In terms of classifers, it was found that the neural network MLP outperforms the continuous Hidden Markov Model with an accuracy of 91.91%, and additional performance requires better feature sets and more training data. Improved EEG features based on time frequency representation were optimized to differentiate Awake with 93.52% sensitivity and 94.60% specificity, differentiate REM with 96.12% sensitivity and 93.63% specificity, differentiate Stages II and III with 96.81% sensitivity and 89.28% specificity, and differentiate Stages III and IV with 93.60% sensitivity and 90.43% specificity. Due to the limited data set, an example of applying contextual information using a One-Cycle-Duo-Direction model was built and shown to improve EEG features by up to 10%. This level of performance is comparable if not superior to the human scorer accuracy of 88% to 94%. This thesis improved some aspects of sleep staging automation, but due to the limitations on resources, the full potential of these improvements could not be demonstrated. To further develop these improvements, additional data sets customized by sleep staging experts is crucial.

Shape memory alloys (SMAs) are a group of alloys that demonstrate the unique ability of returning back to a previously defined shape or size if subjected to the appropriate thermal procedure. They have been implemented as actuators in a wide range of applications spanning several fields such as robotics, aeronautics, automotive and even in medicine. Several controllers, linear and nonlinear, have been designed to control these actuators. However, controlling these actuators is no simple task as they are highly nonlinear due to the hysteresis inherent in them. In fact, their control depends on two important factors: the thermal conditions they are subjected to and the stress applied to them. The former can be further divided into air flow and ambient temperatures. These thermal conditions determine the amount of power needed to heat the SMA wire. In the SMA data sheets, manufacturers specify what they refer to as the "safe current" which is the maximum current value that can be applied to the SMA wire indefinitely without burning it. However, they specify this current value at room temperature and under certain convection conditions. In the work presented here, the focus was the control of SMA actuators under different ambient temperatures. Thus, in this research, the main goal was to design and implement a controller that will actuate, or contract, the SMA wire in approximately the same amount of time regardless of the ambient temperatures with a fixed load applied to it.

A detailed experimental and theoretical investigation of noise in both current mode and voltage mode amorphous silicon (a-Si) active pixel sensors (APS) has been performed in this study. Both flicker (1/f) and thermal noise are considered. The experimental result in this study emphasizes the computation of the output noise variance, and not the output noise spectrum. This study determines which mode of operation is superior in term of output noise. The current noise power spectral density of a single a-Si TFT is also measured in order to find the suitable model for calculating the flicker noise. This experimental result matches Hooge’s model. The theoretical analysis shows that the voltage mode APS has an advantage over the current mode APS in terms of the flicker noise due to the operation of the readout process. The experimental data are compared to the theoretical analysis and are in good agreement. The results obtained in this study apply equally well to APS circuits made using polycrystalline silicon (poly-Si) and single crystal silicon.

Verbal interaction provides a natural and social-style interaction mode by which robots can communicate with general public who is likely unknowledgeable in robotics. This interaction mechanism is also very important for a broad range of users such as hands/eyes-busy users, motor-impaired users, users with vision impairment and users working in hostile environments. Verbal interaction is very popular in robotics especially in personal assistive robots, which are used to help elderly people and in entertainment robots. Several research endeavors have been assigned to endow the robots with verbal interaction as a high-level faculty. However, the language usages of many of them were simple and may not be considered as full speech dialogue systems providing natural language understanding. In this thesis, we investigate a testbed platform that can be deployed to enable human-robot verbal interaction. The proposed approach encompasses a design pattern-based user interface and a user-independent automatic speech recognizer with a modified grammar module in the context of human-robot interaction. The user interface is used to simulate robots response toward multiple users’ voice commands. The performance of the proposed testbed has been evaluated quantitatively using a set of evaluation metrics such as word correct rate, recognition time and success and false action rates. The conducted experiments show the promising features of the system. The results obtained could be refined even further by training the system for more voice commands and the whole system could be ported to real robotic platforms such as Peoplebot to endow it with natural language understanding.

High-Performance computing is one of the major areas making inroads into the future for large-scale simulation. Applications such as 3D nuclear test, Molecular Dynamics, and Quantum Monte Carlo simulations are now developed on supercomputers using the latest computing technologies. As per the TOP500 supercomputers rating, most of todays supercomputers are now heterogeneous: with massively parallel Graphics Processing Units (GPU) equipped with Multi-core CPU(s) to increase the computational capacity. The Variational Monte Carlo(VMC) method is used in the Many Body Physics to study the ground state properties of a system. The wavefunction depends on some variational parameters, which contain the physics for a better prediction. In general, the variational parameters are chosen to realize some sort of order or broken symmetry such as superconductivity and magnetism. The variational approach is computationally expensive and requires a large number of Markov chains (MCs) to obtain convergence. The MCs exhibit abundant data parallelism and parallelizing across CPU clusters will prove to be expensive and does not scale in proportion to the system size. Hence, this method will be a suitable candidate on a massively parallel Graphics Processing Unit (GPU). In this research, we discuss about the various optimization and parallelization strategies adopted to port the VMC method to a NVIDIA GPU using CUDA. We obtained a speedup of nearly 3.85 X compared to the MPI implementation [4] and a speedup of upto 19 X compared to an object-oriented C++ code.

The Superconducting Nanowire single-photon detector (SNSPD) made with niobium-titanium nitride (NbTiN) thin films fabricated on oxidized silicon substrates are highly promising nanodevices. The SNSPD is an immensely capable infrared single photon detector. When cooled down with liquid helium the device exhibits high detection efficiency, low dark-counts in spite of fast response times, and low timing jitter. For good single-photon sensitivity at telecom wavelengths, picosecod timing resolutions (<100 ps), and high counting frequencies, the SNSPD is first choice compared to the alternatives like indium gallium arsenide avalanche diode detectors, silicon single photon avalanche diode detectors, and other superconducting single photon technology like, transition edge detectors. These SNSPD characteristics make it ideal for long distance quantum key distribution (QKD). Although the exploitation of the constructive interference that occurs at the SiO2/Si interface can boost detector efficiency, efficient packaging and fiber coupling is often a limiting factor in the overall system quantum efficiency (SQE). In an attempt to maximize SQE, we use a controlled expansion alloy for optimal performance at cryogenic temperatures. This nickel-iron alloy has a high relative permeability and, therefore, attenuates electromagnetic interference. Our report focuses on the theoretical and experimental methods used in the characterization of an un-patterned NbTiN superconducting nanowire. We use simulations based on the gathered experimental data to design an optimal SNSPD meander and calculate its expected SQE. We also use simulations to analyze the speed and value added to secure key rates in long distance QKD schemes up to 400 km with down-conversion sources with positively spectrally correlated and decorrelated photon pairs at 1550 nm wavelength.

Non-linear programming has been extensively used in wireless telecommunication systems design. An important criterion in optimization is the minimization of mean square error. This thesis examines two applications: peak to average power ratio (PAPR) reduction in orthogonal frequency division multiplexing (OFDM) systems and wireless airtime traffic estimation. These two applications are both of interests to wireless service providers. PAPR reduction is implemented in the handheld devices and low complexity is a major objective. On the other hand, exact traffic prediction can save a huge cost for wireless service providers by better resource management through off-line operations. <br /><br /> High PAPR is one of the major disadvantages of OFDM system which is resulted from large envelope fluctuation of the signal. Our proposed technique to reduce the PAPR is based on constellation shaping that starts with a larger constellation of points, and then the points with higher energy are removed. The constellation shaping algorithm is combined with peak reduction, with extra flexibilities defined to reduce the signal peak. This method, called MMSE-Threshold, has a significant improvement in PAPR reduction with low computational complexity. <br /><br /> The peak reduction formulated into a quadratic minimization problem is subsequently optimized by the semidefinite programming algorithm, and the simulation results show that the PAPR of semidefinite programming algorithm (SDPA) has noticeable improvement over MMSE-Threshold while SDPA has higher complexity. Results are also presented for the PAPR minimization by applying optimization techniques such as hill climbing and simulated annealing. The simulation results indicate that for a small number of sub-carriers, both hill climbing and simulated annealing result in a significant improvement in PAPR reduction, while their degree of complexity can be very large. <br /><br /> The second application of non-linear optimization is in airtime data traffic estimation. This is a crucial problem in many organizations and plays a significant role in resource management of the company. Even a small improvement in the data prediction can save a huge cost for the organization. Our proposed method is based on the definition of extra parameters for the basic structural model. In the proposed technique, a novel search method that combines the maximum likelihood estimation with mean absolute percentage error of the estimated data is presented. Simulated results indicate a substantial improvement in the proposed technique over that of the basic structural model and seasonal autoregressive integrated moving average (SARIMA) package. In addition, this model is capable of updating the parameters when new data become available.

This thesis presents a study of the bias-induced threshold voltage metastability phenomenon of the hydrogenated amorphous silicon (a-Si:H) thin film transistors (TFTs). An application of gate bias stress shifts the threshold voltage of a TFT. After the bias stress is removed, the threshold voltage eventually returns to its original value. The underlying physical mechanisms for the shift in threshold voltage during the application of the bias and after the removal of the bias stress are investigated. <br /><br /> The creation of extra defect states in the band gap of a-Si:H close to the gate dielectric interface, and the charge trapping in the silicon nitride (SiN) gate dielectric are the most commonly considered instability mechanisms of threshold voltage. In the first part of this work, the defect state creation mechanism is reviewed and the kinetics of the charge trapping in the SiN is modelled assuming a simplified mono-energetic and a more realistic Gaussian distribution of the SiN traps. The charge trapping in the mono-energetic SiN traps was approximated by a logarithmic function of time. However, the charge trapping with a Gaussian distribution of SiN traps results in a more complex behavior. <br /><br /> The change in the threshold voltage of a TFT after the gate bias has been removed is referred to threshold voltage relaxation, and it is investigated in the second part of this work. A study of the threshold voltage relaxation sheds more light on the metastability mechanisms of a-Si:H TFTs. Possible mechanisms considered for the relaxation of threshold voltage are the annealing of the extra defect states and the charge de-trapping from the SiN gate dielectric. The kinetics of the charge de-trapping from a mono-energetic and a Gaussian distribution of the SiN traps are analytically modelled. It is shown that the defect state annealing mechanisms cannot explain the observed threshold voltage relaxation, but a study of the kinetics of charge de-trapping helps to bring about a very good agreement with the experimentally obtained results. Using the experimentally measured threshold voltage relaxation results, a Gaussian distribution of gap states is extracted for the SiN. This explains the threshold voltage relaxation of TFT after the bias stress with voltages as high as 50V is removed. <br /><br /> Finally, the results obtained from the threshold voltage relaxation make it possible to calculate the total charge trapped in the SiN and to quantitatively distinguish between the charge trapping mechanism and the defect state creation mechanisms. In conclusion, for the TFTs used in this thesis, the charge trapping in the SiN gate dielectric is shown to be the dominant threshold voltage metastability mechanism caused in short bias stress times.