For wireless network systems, iterative power control algorithms have been proposed to minimize the transmission power, while maintaining reliable communication and base stations. However, since the measurements are random, the channel characteristics always are described by Stochastic Differential Equations (SDE). Based on the stochastic approximation methods, and using time-varying step size sequences, we can get an approximation algorithm to reach an optimal power allocation. After the study of optimal power allocation, the probabilistic Quality of Service (QoS) measures are introduced to evaluate the performance of any control strategy. It provides tight bounds that relate to the probability of failure in achieving the desired QoS requirements. This thesis addresses mobile systems consisting of M transmitters and M receivers, which are subject to motion, and their power is described by SDE. The optimal power control problem is formulated, and the outage probability corresponding to a desired QoS requirements is computed using Moment Generating Function (MGF). Numerical results show that each user needs only to know its own channel gain and its own output assigned by the base station to update the transmitter power in order to maintain a desired Signal to Interference Ratio (SIR) and QoS requirement at the receiver.

This work focuses on the technological environment of M-Commerce, and highlights some technological reasons slowing the penetration of this technology in North America1. The purpose is to provide some closure and present a potential solution that would improve the North American M-Commerce adoption level. The determined research hypotheses are: (1) Lack of adequate government regulations; (2) Lack of adequate spectrum; (3) Lack of adequate technologies; (4) Lack of adequate standardization process. An analysis and a comparison of these variables in the respective poles---Western Europe, Japan, North America---using their respective technology---GSM, I-Mode, CDMA is conducted. More specifically this in-depth comparative analysis helps determine if and how these variables impact the QoS2 and therefore the M-Commerce adoption outcome. The lack of adequate spectrum and the lack of adequate standardization process, in North America show to be the primary reasons for the slow adoption of M-Commerce. Hence, the conclusion of this work proposes an alternative solution. This proposal provides a new revised model of the North American M-Commerce value chain. This model would eliminate the identified issues and increase the Quality of Service, as well as the customer's satisfaction, resulting in higher adoption level. Furthermore, this potential new strategy gives North America the opportunity to reposition itself. 1North America: We define North America by Canada and The United States. However throughout this work we often used cases and examples related to the United States. The reason being, that the United States is leading the wireless telecommunication in North America. 2QoS: Quality of Service

Nowadays, common indices act mainly as indicators of stock market performance and development. Since almost all of these indices are somehow a weighted average, thus, they are not able to answer the main question of investors, which is the likelihood that the status of a market changes or remains the same in next sessions. Therefore, by applying the principles of information theory we will try to suggest a new defined index as a new approach to stock market indicators. In this thesis, we suggest two different ways of market analysis; first, we consider a single stock option and second, we consider a family of stock options. In the first case, we mainly consider three type of information measures; entropy, entropy rate and relative entropy. In the second case, we consider two information measures; entropy rate and joint entropy. We also suggest several stochastic models as an estimate for stock price such as the Geometric Brownian motion model, with constant drift and volatility or with drift as a function of time, and Markov chain models. In addition, the analysis is carried by using models for the option price and the "rate of change" of the option price.

In this thesis, design of the 8-channel DCE (Dynamic Channel Equalizer) for use in WDM (Wavelength Division Multiplexing) systems in general and in particular its adaptation as a "Smart Amplifier Solution" is presented. The innovative Optimized C-band Gain Flattened EDFA (Erbium Doped Fiber Amplifier) design without GFF (Gain Flattening Filter) and its integration with the DCE and embedded control electronics into a Smart Amplifier is proposed. Further improvement of the Smart Amplifier is suggested. The systems design approach, where all subsystems are taken into account to optimize the WDM network was adopted. Hence, the Post and In-Line Amplifiers with APC (Automatic Power Control) are also considered and designed accordingly to achieve the optimized amplification and equalization results for the dynamic WDM networks. The experimental results and discussion for each design are presented and simulation data are provided for comparison and confirmation of the proposed Smart Amplifier design. In addition, simulation results with respect to feasibility of the Optimized C+L band EDFA and conditions under which it can be used as a broadband ASE source are discussed.

In this thesis, we construct two new algorithms for image denoising, namely, spline and spline-wavelet, which combine spline interpolation and wavelets together with nonlinear filtering based on block singular value decomposition. Those two approaches are compared with other existing methods, which involve BlockSvd filter, wavelet (global thresholding) filter, median filter, average filter, and adaptive filter. The performance of these approaches differs little from each other. Generally speaking, median filter is very suitable for processing images to reduce "salt and pepper" noise. But for zero-mean Gaussian and speckle noises, an adaptive filter and spline-wavelet methods are more stable and slightly superior to other filters in most conditions and for most images. The proposed algorithms were tested under different types of images and a wide range of signal to noise ratios (SNR). The numerical results demonstrate that these methods can be used in different and useful ways for reducing image noise.

The widespread use of the Internet signals the need for a better understanding of trust as a basis for secure on-line interaction. In the face of increasing uncertainty and risk, users and machines must be allowed to reason effectively about the trustworthiness of other entities. In this thesis, we propose a trust model that assists users and machines with decision-making in online interactions by using past behavior as a predictor of likely future behavior. We develop a general method to automatically compute trust based on self-experience and the recommendations of others. Our trust model solves the problem of recommendation combination and detection of unfair recommendations. Our approach involves data analysis methods (Bayesian estimation, Dirichlet distribution), and machine learning methods (Weighted Majority Algorithm). Furthermore, we apply our trust model to several utility models to increase the accuracy of decision-making in different contexts of Web Services. We describe simulation experiments to illustrate its effectiveness, robustness and the evolution of trust.

In this thesis, we present a real time feedback control strategy to optimize the dynamic performance of computer communication network. In previous studies [8-10], feedback delay, arising from communication delay, was shown to degrade system performance. Considering this negative impact of delay, we propose a new control law which predicts, in advance, the traffic and exercises control based on the predicted traffic. In experiments, we apple the token bucket (TB) mechanism to construct a discrete dynamic system model [8-10], in which one multiplexor, linked to all the TBs, multiplexes the conforming traffic that have been policed at TBs. We demonstrate that the improvement of the system performance by presenting the simulation results corresponding to different stochastic traffic models. The experiments and analysis in this thesis provide valuable insight for the network researchers to do traffic optimal control [35].

Background. The Decisional Conflict Scale (DCS) is a useful clinical tool in assessing the health care consumers' decisional needs. However, the exact predictive ability of the DCS is unknown. Objective. To examine the DCS as a proxy measure for patients' knowledge level and as a predictor of downstream effects of high decisional conflict, specifically, decision delay, discontinuance of chosen option, and decisional regret. Design. Meta-analyses with individual data from 10 clinical trials using the following statistics: descriptive statistics, correlation analysis and logistic regression. Results. Patients' knowledge deficit has a fair association with the uninformed subscale of the DCS (OR 3.10; 95% CI 1.58-6.05). Patients' decisional delay has a very strong association with the DCS (OR 23.81; 4.66-121.51). Patients' discontinuance of chosen treatment has a varied association with the DCS, very strong for change from status quo (OR 59.37; 4.09-861.05) and fair for change from active treatment (OR 3.39; 1.42-8.00). Patients' decisional regret has a strong association with the DCS (OR 5.52; 3.35-9.12). Conclusion. When clinicians assess a patient's decisional conflict after counselling, and they find the decisional conflict is low, they can be reasonably assured that the likelihood of downstream decision delay, change from the status quo, or decisional regret will be low. However, low scores on the uninformed scale do not guarantee the patient is well informed; they need to validate the patients' understanding with some follow-up questions. Moreover, the likelihood of discontinuing active treatment even with low decisional conflict is also a possibility.

This research is to propose a Web services based framework for e-learning portal systems designed to present a flexible integration model and to make the e-learning environment intelligent and adaptive to each individual learner. It is very important that all the learning components and applications are clearly defined, effectively discovered and loosely coupled. Web services and Portals provide an essential deployment environment to realize collaborative e-learning environment by facilitating efficient communication of components and applications. Given the proposed framework, instructor and learner will be able to use their learning objects or services based on portal (collaboration environment) universally anywhere, any time through common communication protocols (HTTP, SOAP). The key values of interoperability and accessibility in the proposed architecture enhance the future collaborative e-learning portal systems to communicate more efficiently and to share data more easily. A prototype of the proposed system is designed and implemented in a J2EE (Java 2 Enterprise Edition) combined with multi-channel, using WSRP (Web Service for Remote Portlet) for building essential Web services and for parsing SOAP (Simple Object Access Protocol) messages on lightweight platforms. The implementation is a successful demonstration that learning services can be easily accessed by portlet through standard Web services interface. Another implementation about role base access control is presented. An internationalization solution (I18N, L10N) based web-service e-learning portal system is successfully accomplished in this implementation.

With the high increase in the availability of digital music, it has become of interest to automatically query a database of musical pieces. At the same time, a feasible solution of this objective gives us an insight into how humans perceive and classify music. In this research, we discuss our approach to classify music into four categories: pop, classical, country and jazz. Songs are collected in wave format. We randomly chose five 10-second clips from different parts of a song. We discussed two families of features: wavelet features and time-based features. These features are capable of capturing the information of energy and time of voice signal. Instead of using traditional Mel-Frequency Cepstral Coefficients (MFCC)[7] methods, which are widely used in audio classification and music classification, we incorporate the features in statistical classification methods such as LDA, QDA and tree. Finally, we attempted to create an adaptive tree approach for classification. In this research, 130 songs are collected. Pop songs are collected in 4 languages, English, Chinese, Spanish and French. A cross validation method is used to compute the proportion of correctly classified songs. It is shown that the tree method has a proportion of correct classification equal 0.80 when pop and country are considered as one category.