While the RF building blocks of narrowband system-on-chip designs have increasingly been created in CMOS during the past decade, researchers have started to look at the possibility of implementation of broadband transceivers in CMOS technology. High speed optical links with operating frequencies of up to 40 GHz and ultra wideband (UWB) wireless systems operating in 3 to 10 GHz frequency band are examples of these broadband applications. CMOS offers a low fabrication cost, and a higher level of integration compared with compound semiconductor technologies that currently claim broadband RFIC applications. <br /><br /> In this work, we focus on the design of broadband low-noise amplifiers: the fundamental building blocks of high data rate wireline and wireless telecommunication systems. A well established microwave engineering technique -distributed amplification- with a potential bandwidth up to the cut-off frequency of transistors is employed. However, the implementation of distributed amplifiers in CMOS imposes new challenges, such as gain attenuation because of substrate loss of on-chip inductors, a typical large die area, and a large noise-figure. These problems have been addressed in this dissertation as described below. <br /><br /> On-chip inductors, the essential components of the distributed amplifiers' gate and drain transmission lines, dissipate more and more power in silicon substrates as well as in metal lines as frequency increases, which in turn reduces the gain and deteriorates the input/output matching. Using active negative resistors implemented by a capacitively source degenerated configuration, we have fully compensated the loss of the transmission lines in order to achieve a flat gain of 10 dB over the entire DC-to-44 GHz bandwidth. <br /><br /> We have addressed another drawback of distributed amplifiers, large die area, by utilizing closely-placed RF transmission lines instead of spiral inductors. Because of a more compact implementation of transmission lines, the area of the distributed amplifiers is considerably reduced at the expense of extra design steps required for the modeling of the closely-placed RF transmission lines. A post-layout simulation method is developed to take into account the effect of inductive and capacitive coupling by incorporating a 3D EM simulator into the design process. A 9-dB 27-GHz distributed amplifier has been fabricated in an area as small as 0. 17 <em>mm</em>² using 180nm TSMC's CMOS process. <br /><br /> For wireless applications (UWB), a very low-noise figure is required for the broadband preamplifier. Conventional distributed amplifiers fail to provide a low noise figure mainly because of the noise injected by the terminating resistor of the gate transmission lines. We have replaced the terminating resistor with a frequency-dependent resistor which trades off the low frequency input matching of the distributed amplifier (not required for UWB) with a better noise performance. Our proposed design provides a gain of 12 dB with an average noise figure of 3. 4 dB over the entire 3-10 GHz band, advancing the state-of-the-art implementation of broadband LNAs.

With the recent advances in the development of wireless communication networks, Wireless Mesh Networks (WMNs) have been receiving considerable research interests in recent years. Many challenges need to be addressed for successful WMN deployment. One of the fundamental challenges is the need to support integrated services and provision different Quality of Service (QoS) for various applications. In order to allow differentiated services, Medium Access Control (MAC) has to provide priority management techniques at the link layer. In Code Division Multiple Access (CDMA) based WMNs, the interference phenomenon and the simultaneous transmissions must be considered. We propose two priority schemes for a distributed CDMA-based MAC WMNs. We take into account interference, multiple services, QoS requirements for each type of traffic, and the simultaneous transmission in CDMA. The first priority scheme is within a node. Each node has independent queues for each traffic class. According to QoS requirements, the queue that should be served first is determined. The second priority scheme is among neighbour nodes. It is proposed for possible multiple simultaneous transmissions with CDMA. This scheme gives a higher chance of correct transmission to high priority traffic than low priority traffic. In addition, we propose to use an adaptive spreading gain and a frame structure to achieve high resource utilization. Simulation results demonstrate that the proposed schemes can achieve effective QoS guarantee.

The problem of rank deficient multiple input multiple out (MIMO) systems arises when the number of transmit antennas M is greater than number of receive antennas N or when the channel gains are strongly correlated. Most of the optimal algorithms that deal with uncoded rank-deficient (under-determined) V-BLAST MIMO systems (e.g. Damen ,Meraim and Belfiore) suffer from high complexity and large processing time. Recently, some new optimal algorithms were introduced with low complexity for small constellations like 4-QAM yet they still suffer from very high complexity and processing time with large constellations like the 16 QAM. In order to reduce the complexity and the processing time of the decoding algorithms, some suboptimal algorithms were introduced. One of the most efficient suboptimal solutions for this problem is based on the Minimum mean square error decision-feedback equalizer (MMSE-DFE) followed by either sphere decoder or fano decoder. The performance of these algorithms is shown to be a fraction of dB from the maximum likelihood decoders while offering outstanding reduction in complexity compared to the most efficient ML algorithms (e.g. Cui and Tellambura algorithm). These suboptimal algorithms employ a two stage approach. In the first stage, the channel is pre-processed to transform the original decoding problem into a simpler form which facilitates the search decoding step. The second stage is basically the application of the sphere decoding search algorithm in the case of MMSE-DFE sphere decoding step or Fano decoder in the case of MMSE-DFE Fano decoder. In this study, various algorithms which deal with rank deficient MIMO systems such as Damen,Meraim and Belfiore algorithm ,Dayal and Varansi algorithm, and Cui and Tellambura algorithm are discussed and compared. Moreover, the MMSE-DFE sphere decoding algorithm and MMSE-DFE fano decoding algorithm are applied on uncoded V-BLAST rank deficient MIMO systems. The optimality of MMSE-DFE sphere decoding algorithm is analyzed in the case of V-BLAST 4-QAM. Furthermore, Simulation results show that when these algorithms are extended to cover large constellations, their performance falls within a fraction of dB behind the ML while achieving a significant decrease in the processing time by more than an order of magnitude when compared to the least

A gel electrophoresis technology based on absorption of ultraviolet radiation for fragment sizing of bio-molecular segments such as protein and nucleic acid is introduced for the first time. The new technology has the potential to improve conventional gel electrophoresis method by lowering the cost and increasing the throughpxcvfut. A cost effective, high sensitivity, short wavelength selective detector is an essential component to enable the proposed technology. In this thesis, hydrogenated amorphous silicon (a-Si:H) metal semiconductor metal(MSM) are investigated as the short wavelength detector of choice. The operation of planar MSM photoconductor-based photo detectors with very thin a-Si:H film thickness and aluminium electrodes is investigated. Experimental results of photocurrent measurements as well as responsivity and quantum efficiency are presented. The MSM photodetectors presented are fully compatible with state-of-the-art staggered gate thin film transistor (TFT) fabrication processes to enable large area pixel arrays for bio-molecular imaging.

An all-IP pervasive networking system provides a comprehensive IP solution where voice, data and streamed multimedia can be delivered to users at anytime and anywhere. Network selection is a key issue in this converged heterogeneous networking environment. A traditional way to select a target network is only based on the received signal strength (RSS); however, it is not comprehensive enough to meet the various demands of different multimedia applications and different users. Though some existing schemes have considered multiple criteria (e.g. QoS, security, connection cost, etc.) for access network selection, there are still several problems unsettled or not being solved perfectly. In this thesis, we propose a novel model to handle this network selection issue. Firstly, we take advantage of IEEE 802.21 to obtain the information of neighboring networks and then classify the information into two categories: 1) compensatory information and 2) non-compensatory information; secondly, we use the non-compensatory information to sort out the capable networks as candidates. If a neighboring network satisfies all the requirements of non-compensatory criteria, the checking of the compensatory information will then be triggered; thirdly, taking the values of compensatory information as input, we propose a hybrid ANP and RTOPSIS model to rank the candidate networks. ANP elicit weights to compensatory criteria and eliminates the interdependence impact on them, and RTOPSIS resolves the rank reversal problem which happens in some multiple criteria decision making (MCDM) algorithms such as AHP, TOPSIS, and ELECTRE. The evaluation study verifies the usability and validity of our proposed network selection method. Furthermore, a comparison study with a TOPSIS based algorithm shows the advantage and superiority of the proposed RTOPSIS based model.

In this thesis, I mainly focus on the evaluation of the upper bounds to the capacity of wireless networks. With the consideration of the two measures, the maximal transmission rate for any source-destination pair and the transport capacity of wireless networks, I summarize the most recent results to the upper bounds of these two measures first in this thesis. At the same time, I also improve and modify the previous results given in these papers. Moreover, I present a proof to the upper bound of maximal transmission rate with high probability by taking the fading of the channel into account when the full CSI is only known to the receivers. With a simple extension of the result, I derive an upper bound to the transport capacity of wireless networks without full CSI at the receiver side. A linear scaling of the upper bound to transport capacity is also derived when the path loss exponent is greater than three. Compared with the previous results, it is shown that the upper bound given in this thesis is much better for relatively large alpha and a minimum distance constraint.

The closed-set speaker identification problem is defined as the search within a set of persons for the speaker of a certain utterance. It is reported that the Gaussian mixture model (GMM) classifier achieves very high classification accuracies (in the range 95% - 100%) when both the training and testing utterances are recorded in sound proof studio, i.e., there is neither additive noise nor spectral distortion to the speech signals. However, in real life applications, speech is usually corrupted by noise and band-limitation. Moreover, there is a mismatch between the recording conditions of the training and testing environments. As a result, the classification accuracy of GMM-based systems deteriorates significantly. In this thesis, we propose a two-step procedure for improving the speaker identification performance under noisy environment. In the first step, we introduce a new classifier: vector autoregressive Gaussian mixture (VARGM) model. Unlike the GMM, the new classifier models correlations between successive feature vectors. We also integrate the proposed method into the framework of the universal background model (UBM). In addition, we develop the learning procedure according to the maximum likelihood (ML) criterion. Based on a thorough experimental evaluation, the proposed method achieves an improvement of 3 to 5% in the identification accuracy. In the second step, we propose a new compensation technique based on the generalized maximum likelihood (GML) decision rule. In particular, we assume a general form for the distribution of the noise-corrupted utterances, which contains two types of parameters: clean speech-related parameters and noise-related parameters. While the clean speech related parameters are estimated during the training phase, the noise related parameters are estimated from the corrupted speech in the testing phase. We applied the proposed method to utterances of 50 speakers selected from the TIMIT database, artificially corrupted by convolutive and additive noise. The signal to noise ratio (SNR) varies from 0 to 20 dB. Simulation results reveal that the proposed method achieves good robustness against variation in the SNR. For utterances corrupted by covolutive noise, the improvement in the classification accuracy ranges from 70% for SNR = 0 dB to around 4% for SNR = 10dB, compared to the standard ML decision rule. For utterances corrupted by additive noise, the improvement in the classification accuracy ranges from 1% to 10% for SNRs ranging from 0 to 20 dB. The proposed VARGM classifier is also applied to the speech emotion classification problem. In particular, we use the Berlin emotional speech database to validate the classification performance of the proposed VARGM classifier. The proposed technique provides a classification accuracy of 76% versus 71% for the hidden Markov model, 67% for the k-nearest neighbors, 55% for feed-forward neural networks. The model gives also better discrimination between high-arousal emotions (joy, anger, fear), low arousal emotions (sadness, boredom), and neutral emotions than the HMM. Another interesting application of the VARGM model is the blind equalization of multi input multiple output (MIMO) communication channels. Based on VARGM modeling of MIMO channels, we propose a four-step equalization procedure. First, the received data vectors are fitted into a VARGM model using the expectation maximization (EM) algorithm. The constructed VARGM model is then used to filter the received data. A Baysian decision rule is then applied to identify the transmitted symbols up to a permutation and phase ambiguities, which are finally resolved using a small training sequence. Moreover, we propose a fast and easily implementable model order selection technique. The new equalization algorithm is compared to the whitening method and found to provide less symbol error probability. The proposed technique is also applied to frequency-flat slow fading channels and found to provide a more accurate estimate of the channel response than that provided by the blind de-convolution exploiting channel encoding (BDCC) method and at a higher information rate.

We propose an information theoretic framework for scheduling the transmissions in a two-way multi-hop network. First we investigate some long standing open problems that were encountered during the course of the research. To illustrate their difficulty, we describe some failed attempts at resolving them. We then introduce the two-way one-relay channel. It turns out that the achievable rate region of this network has a nice interpretation, especially when viewed in the context of the open problems examined earlier. Motivated by this observation, we attempt to extend this achievable region to the two-way two-relay channel. In the process, we expose a fundamental deadlock problem in which each relay needs to decode before the other in order to enable mutual assistance. Our most important contribution is a resolution to this deadlock problem; we add an additional constraint that ensures some relay can decode at least one message before the other relay. Furthermore, we also introduce several coding schemes to prove that the additional constraint is indeed sufficient. Our schemes also show that information theory provides unique insight into scheduling the transmissions of multi-hop networks.

To support real-time applications, we present a Measurement-based Admission Control (MBAC) scheme with Modified Largest Weighted Delay First (M-LWDF) scheduling algorithm. The objective of the admission control scheme is to admit new real-time application call into the system without jeopardizing the maximum average packet delay bound. Measured values of the average packet delay from the network are used for the admission decision. As long as a new call can obtain the requested service and the packet delay of existing calls are not risked by admitting it, the new call will be accepted into the network. In addition, M-LWDF scheduling algorithm is introduced to efficiently allocate network resource. Simulation results show that the proposed MBAC scheme maintains good packet delay bound.

A speaker's accent is the most important factor affecting the performance of Natural Language Call Routing (NLCR) systems because accents vary widely, even within the same country or community. This variation also occurs when non-native speakers start to learn a second language, the substitution of native language phonology being a common process. Such substitution leads to fuzziness between the phoneme boundaries and phoneme classes, which reduces out-of-class variations, and increases the similarities between the different sets of phonemes. Thus, this fuzziness is the main cause of reduced NLCR system performance. The main requirement for commercial enterprises using an NLCR system is to have a robust NLCR system that provides call understanding and routing to appropriate destinations. The chief motivation for this present work is to develop an NLCR system that eliminates multilayered menus and employs a sophisticated speaker accent-based automated voice response system around the clock. Currently, NLCRs are not fully equipped with accent classification capability. Our main objective is to develop both speaker-independent and speaker-dependent accent classification systems that understand a caller's query, classify the caller's accent, and route the call to the acoustic model that has been thoroughly trained on a database of speech utterances recorded by such speakers. In the field of accent classification, the dominant approaches are the Gaussian Mixture Model (GMM) and Hidden Markov Model (HMM). Of the two, GMM is the most widely implemented for accent classification. However, GMM performance depends on the initial partitions and number of Gaussian mixtures, both of which can reduce performance if poorly chosen. To overcome these shortcomings, we propose a speaker-independent accent classification system based on a distance metric learning approach and evolution strategy. This approach depends on side information from dissimilar pairs of accent groups to transfer data points to a new feature space where the Euclidean distances between similar and dissimilar points are at their minimum and maximum, respectively. Finally, a Non-dominated Sorting Evolution Strategy (NSES)-based k-means clustering algorithm is employed on the training data set processed by the distance metric learning approach. The main objectives of the NSES-based k-means approach are to find the cluster centroids as well as the optimal number of clusters for a GMM classifier. In the case of a speaker-dependent application, a new method is proposed based on the fuzzy canonical correlation analysis to find appropriate Gaussian mixtures for a GMM-based accent classification system. In our proposed method, we implement a fuzzy clustering approach to minimize the within-group sum-of-square-error and canonical correlation analysis to maximize the correlation between the speech feature vectors and cluster centroids. We conducted a number of experiments using the TIMIT database, the speech accent archive, and the foreign accent English databases for evaluating the performance of speaker-independent and speaker-dependent applications. Assessment of the applications and analysis shows that our proposed methodologies outperform the HMM, GMM, vector quantization GMM, and radial basis neural networks.