The changing nature of Internet-based applications is imposing stricter demands on the performance of the Internet. For many new applications requiring differentiated quality of service, the best effort model of the internet is no longer adequate. Multiprotocol label switching (MPLS) overlays a connection oriented network on the connection-less IP networks, thereby addressing several shortcomings of the IP network. MPLS improves the forwarding speed, and it provides a virtual path capability to efficiently carry differentiated services. Additionally, MPLS enables traffic engineering by explicit traffic engineering tunnels to be set up across the network to utilize all available bandwidth in an efficient manner. Constraint-based routing has been proposed as an effective approach to implement traffic engineering in MPLS. In this thesis, we propose several constraint-based routing algorithms to support both Quality of Service (QoS) and BE traffic of VPNs over the MPLS network. Our algorithms also show the explicit route setup for VPNs by including attributes like node affinity and link affinity constraints. We also study the path length limit for the QoS traffic of VPNs which may have stringent hop constraints due to delay requirements. Finally, we propose various path protection schemes for MPLS VPNs. Our approach for VPNs are scalable to large networks.

The recent years brought a phenomenal development of Internet. It is, therefore, important to find some ways to improve the performances. The first step in this direction is the characterization and modeling of the network traffic. It has been tested that the network traffic behaves like a self-similar process, while packets interarrivals time possess the long-range dependence property. In particular, we model them by using fractional Brownian motion and fractional Gaussian noise, respectively. Note that, the former is just the cumulative sum of the latter. By using these concepts, the traffic characterization reduces to the estimation of one value: the Hurst parameter. Numerous methods exist to evaluate this parameter. Nevertheless, a few studies take account of the inherent non-stationarity present in real data. For short samples, the stationarity hypothesis might hold. But for larger samples, this is hardly the case. As an example, for network traffic, the day cycle shows non-stationarity. By not considering the non-stationarity, an inaccurate or even inappropriate estimation may result. Our objective in this thesis is to test the robustness of several techniques such as aggregated variance method, rescaled range method, and wavelets method, in presence of a set of non-stationarity trends. We study the estimators on a known signal generated using Hosking, Davies and Harte method, or wavelets-based synthesis. We add various deterministic non-stationarity trends to the original signal. We considered polynomial, power-law, sinusoidal, and level-shift trends. Results help analyze the behavior of the estimators. All the simulations are carried out using Matlab. We show that, depending on the trend, the estimators react differently. We have also used real data to verify the effectiveness of estimators. Results confirm the observations that we have made with lab data. In particular, we show that the wavelets method provides several flaws. Especially, its results must be carefully analyzed when the data is non-stationary.

This thesis improves upon the work carried out by Mandhani on decimal sequence watermarking showing that the embedding and decoding algorithms used by Mandhani are not robust. We next specify improvements to these algorithms so that the peak signal to noise ratio (PSNR) is maximized and the distortion introduced in the image due to the embedding is minimized. By exploiting the cross correlation property of decimal sequences, the concept of embedding more than one watermark in the same cover image has been investigated. A comparison between the PSNRs for decimal sequence watermarking and other pseudorandom sequences has been made.

This dissertation focuses on providing solutions to two of the most important problems in wireless communication systems design, namely, 1) the interference suppression, and 2) the channel parameter estimation in wireless communication systems over time-varying multipath fading channels. We first study the interference suppression problem in various communication systems under a unified multirate transmultiplexer model. A state-space approach that achieves the optimal realizable equalization (suppression of inter-symbol interference) is proposed, where the Kalman filter is applied to obtain the minimum mean squared error estimate of the transmitted symbols. The properties of the optimal realizable equalizer are analyzed. Its relations with the conventional equalization methods are studied. We show that, although in general a Kalman filter has an infinite impulse response, the Kalman filter based decision-feedback equalizer (Kalman DFE) is a finite length filter. We also propose a novel successive interference cancellation (SIC) scheme to suppress the inter-channel interference encountered in multi-input multi-output systems. Based on spatial filtering theory, the SIC scheme is again converted to a Kalman filtering problem. Combining the Kalman DFE and the SIC scheme in series, the resultant two-stage receiver achieves optimal realizable interference suppression. Our results are the most general ever obtained, and can be applied to any linear channels that have a state-space realization, including time-invariant, time-varying, finite impulse response, and infinite impulse response channels. The second half of the dissertation devotes to the parameter estimation and tracking of single-input single-output time-varying multipath channels. We propose a novel method that can blindly estimate the channel second order statistics (SOS). We establish the channel SOS identifiability condition and propose novel precoder structures that guarantee the blind estimation of the channel SOS and achieve diversities. The estimated channel SOS can then be fit into a low order autoregressive (AR) model characterizing the time evolution of the channel impulse response. Based on this AR model, a new approach to time-varying multipath channel tracking is proposed.

With the continuously expanding of market for portable devices such as wireless communication devices, portable computers, consumer electronics and implantable medical devices, low power is becoming increasingly important in integrated circuits. The low power design can increase operation time and/or utilize a smaller size and lighter-weight battery. In this dissertation, several low power complementary metal-oxide-semiconductor (CMOS) integrated circuit design techniques are investigated. A metal-oxide-semiconductor field effect transistor (MOSFET) can be operated at a lower voltage by forward-biasing the source-substrate junction. This approach has been investigated in detail and used to designing an ultra-low power CMOS operational amplifier for operation at ± 0.4 V. The issue of CMOS latchup and noise has been investigated in detail because of the forward biasing of the substrates of MOSFETs in CMOS. With increasing forward body-bias, the leakage current increases significantly. Dynamic threshold MOSFET (DTMOS) technique is proposed to overcome the drawback which is inherent in a forward-biased MOSFET. By using the DTMOS method with the forward source-body biased MOSFET, two low-power low-voltage CMOS VLSI circuits that of a CMOS analog multiplexer and a Schmitt trigger circuits are designed. In this dissertation, an adaptive body-bias technique is proposed. Adaptive body-bias voltage is generated for several operational frequencies. Another issue, which the chip design community is facing, is the development of portable, cost effective and low power supply voltage. This dissertation proposes a new cost-effective DC/DC converter design in standard 1.5 um n-well CMOS, which adopts a delay-line controller for voltage regulation.

Markov decision processes have become an indispensable tool in applications as diverse as equipment maintenance, manufacturing systems, inventory control, queuing networks and investment analysis. Typically we have a controlled Markov chain on a suitable state space in which transitional probabilities depend on the policy (or decision maker) which comes from a set of possible actions. The main problem of interest would be to find an optimal policy that minimizes the associated cost. Linear Programming has been widely used to find the optimal Markov decision policy. It requires solutions of large systems of simultaneous linear equations. By the fact that the complexity in linear programming increases much faster with the increase in the number of states which is often called curse of dimensionality, the linear programming method can handle only small models. This thesis presents a new method to lessen the curse of dimensionality. By assuming certain monotonicity property for the transition probability, it is shown that a fuzzy membership function can be used to reduce the number of states. The use of membership functions help to reduce the number of the states. However all the states remain intact through the use of the membership value. That is, those states eliminated can be recovered through interpolation with the aid of membership functions. This new proposed method is shown to be effective in coping with the curse of dimensionality.

The design and principle of operation, as well as the electromechanical phenomena of a single-phase linear reluctance motor are discussed. The motor with transverse magnetic flux consists of a primary part, which is moving and a secondary part which is stationary and does not have any windings. The motor can operate under AC or DC supply. When supplied from an AC source it must be equipped with a capacitor connected in series with the coil. In this case the motor operates on the basis of resonance in an RLC primary circuit. When supplied from a DC source it must be equipped with a controlled switch connected to the primary circuit. In this case it operates as a linear switched reluctance motor. A comparison of the motor performance operating under AC and DC supply is presented. The objectives of the project were to design the motor and to determine its performance under AC and DC supply. Design calculations were focused on determining the resistance, the inductance and the mass of the primary part. The calculations of primary winding inductance and magnetic flux density distribution were performed using finite element method. In order to determine the motor performance the simulation of motor operation under AC and DC supply was carried out using MATLAB/SIMULINK software package. For this purpose the mathematical models of the motor were defined and block diagrams were built. The simulation results presented in this thesis show a better performance of the motor when supplied from DC source. The maximum efficiency that could be obtained is 55%. A study of the influence of the switching angle on the motors electromechanical characteristics shows that the motor performs better when switched ON earlier before the motor develops the positive driving force.

Traffic Engineering (TE) attempts to establish paths for the flow of data in a network so as to optimize the resource utilization and maximize the network performance. One of the main goals of Traffic Engineering is to bring about efficient and reliable network operations. Constraint based routing is the key to Multiprotocol label switching TE. Constraint based routing helps to manage traffic paths within a MPLS network and allows for traffic to flow along certain desired paths. MPLS also supports explicit routing thus providing TE capabilities. MPLS TE enables resiliency and reliability to be built into networks, thus increasing the value and availability of the network. TE is deployed in MPLS networks by providing TE extensions to interior gateway protocols like OSPF and IS-IS. MPLS creates a connection oriented model over the traditional connectionless framework of IP based networks. This connection oriented model helps to overcome several shortcomings of the IP network and also provide the necessary framework to give quality guarantees to IP traffic. Quality of Service (QoS) can be built into MPLS networks where TE is used. MPLS has made significant progress in recent years and is used for deployment of networks all over the world. In this thesis, we propose several constraint-based routing algorithms for MPLS based VPN's. Our algorithms support routing of both Quality of Service (QoS) and BE traffic over the MPLS network. We have implemented a route based and a link based approach for routing of QoS and BE traffic. We also implement the 1 + 1 and 1 : 1 link disjoint and node disjoint path protection mechanisms for the QoS traffic. We study the effect of the various parameters used in our equations such as epsilon, granularity, earning rate and hop count. The problem is formulated as a multicommodity flow (MCF) problem and is solved using the optimization tool - ILOG CPLEX. Our approach is scalable to large networks having a number of VPN's.

The last few years have witnessed an ever growing interest in the field of mobile location systems for cellular systems. The motivation is the series of regulations passed by Federal Communications Commission, requiring that wireless service providers support a mobile telephone callback feature and cell site location mechanism. A further application of the location technology is in the rapidly emerging field of intelligent transportation systems, which are intended to enhance highway safety, location based billing etc. Many of the existing location technologies use GPS and its derivatives which require a specialized subscriber equipment. This is not feasible for popular use, as the cost of such equipments is very high. Hence, for a CDMA network, various methods have been studied that use the cellular network as the sole means to locate the mobile station (MS), where the estimates are derived from the signal transmitted by the MS to a set of base station's (BS) This approach has the advantage of requiring no modifications to the subscriber equipment. While subscriber location has been previously studied for CDMA networks, the effect of multiple access interference has been ignored. In this thesis we investigate the problem of subscriber location in the presence of multiple access interference. Using MATLAB as a simulation tool, we have developed an extensive simulation technique which measures the error in location estimation for different network and user configurations. In our studies we include the effects of log-normal shadow and Rayleigh fading. We present results that illustrate the effects of varying shadowing losses, number of BS's involved in position location, early-late discriminator offset and cell sizes in conjunction with the varying number of users per cell on the accuracy of radiolocation estimation.

The overall aim of this thesis is to deposit carbon layers by thermal Chemical Vapor Deposition (CVD) and investigate the effects of process parameters on the morphology and structure of the deposited carbon layers. Cyclohexane (C6H12) hydrocarbon is the precursor material used as the source of carbon. A 20% Cu- 80% Ni alloy, deposited by RF magnetron sputtering and annealed at the growth temperature either in an argon flow or in 10% hydrogen-90% argon flow, is used as a catalyst prior to carbon film deposition by CVD. The CVD temperature is varied between 550 0C and 800 0C. The Cu-Ni film thickness for CVD is varied from 2 nm to 80 nm. Carbon layers obtained after each CVD growth run of 12 min are characterized by scanning electron microscopy (SEM). The most likely temperature for carbon nanotube (CNT) growth seems to be 725 0C. Growth of fibers and rods is found to occur on CuNi films deposited on oxide coated silicon substrates and not on silicon substrates. The Cu-Ni film on oxide should not be more than 15 nm thick for argon anneal to produce CNTs upto 150 nm in diameter. Cu-Ni catalytic films of thicknesses 6 nm, 12 nm and 24 nm on patterned oxidized substrates are annealed in Ar and Ar-H2 flows for 15 and 30 min at 760 Torr. Smallest size catalyst particles are observed on oxide coated substrates. CVD growth is carried out on the 6 nm and 12 nm catalytic Cu-Ni films for 12 min on oxide coated substrates at 725 0C in Ar-H2 atmosphere. In the first case, 50 nm diameter carbon fibers are produced while 160 nm diameter carbon fibers are observed for the 12 nm thick Cu-Ni film. Transmission electron microscopy (TEM) analysis on the carbon fibers revealed multi-walled structures, 50 nm and 130 nm in diameters indicating the growth of CNTs comparable to the average size of the catalyst particles after annealing. Resistivity measurements on the carbon layers on 6 nm and 12 nm Cu-Ni films show sheet resistivity values of 293 §Ù/square and 85 §Ù/square respectively.