Approaching parameter estimation from the discrete-time domain is the dominating paradigm in system identification. Identification of continuous-time models on the other hand is motivated by the fact that modelling of physical systems often take place in continuous-time. For many practical applications there is also a genuine interest in the parameters connected to these physical models. The most important element of time- and frequency-domain identification from sampled data is the discrete-time system, which is connected to the parameters of the underlying continuous-time system. For input-output models, it governs the frequency response from the sampled input to the sampled output. In case of time series, it models the spectrum of the sampled output. As the rate of sampling increase, the relationship between the discrete- and continuous-time parameters can become more or less ill-conditioned. Mainly, because the gathering of the poles of the discrete-time system around the value 1 in the complex plane will produce numerical difficulties while mapping back to the continuous-time parameters. We will therefore investigate robust alternatives to using the exact discrete-time system, which are based on more direct use of the continuous-time system. Another, maybe more important, reason for studying such approximations is that they will provide insight into how one can deal with non-uniformly sampled data. An equally important issue in system identification is the effect of model choice. The user might not know a lot about the system to begin with. Often, the model will only capture a particular aspect of the data which the user is interested in. Deviations can, for instance, be due to mis-readings while taking measurements or un-modelled dynamics in the case of dynamical systems. They can also be caused by misunderstandings about the continuous-time signal that underlies sampled data. From a user perspective, it is important to be able to control how and to what extent these un-modelled aspects influence the quality of the intended model. The classical way of reducing the effect of modelling errors in statistics, signal processing and identification in the time-domain is to introduce a robust norm into the criterion function of the method. The thesis contains results which quantify the effect of broad-band disturbances on the quality of frequency-domain parameter estimates. It also contains methods to reduce the effect of narrow-band disturbances or frequency domain outliers on frequency-domain parameter estimates by means of methods from robust statistics.

No description available.

Fiber tapers have attracted much attention and have been successfully employed in various applications, ranging from resonators, filters, interferometers to sensors. This thesis studies the properties of fiber tapers for the purpose of making tapered-based devices and sensors in aerospace related application where small size and light weight are critical. This thesis includes theoretical derivation and experimental verifications of distributed mode coupling in tapered single-mode fibers (SMFs) with high-resolution optical frequency-domain reflectometry (OFDR) technique. The studies are realized with OFDR through phase detection of a Mach-Zehnder interferometer (MZI), which measures local refractive index change relative to the reference arm. The wavelength shifts converted by the phase change give the group index differences between the fundamental mode and higher-order modes of fiber tapers. The energy re-distribution is observed in Rayleigh backscatter amplitude as a function of fiber length with a ~13µm resolution over the entire fiber taper, and group index difference between core and cladding modes is measured with a spatial resolution of ~2cm by using autocorrelation data processing. The thermal and mechanical properties of fiber tapers have also been characterized with OFDR. The cross-correlation wavelength shift is related to the refractive index change of the modes. It is shown that residual stress induced by the tapering process results in the inhomogeneous thermal property, which can be significantly reduced by an annealing treatment. A fiber taper with a waist diameter of ~6µm has a force sensitivity of ~620.83nm/N, ~500 times higher than that of SMF. Furthermore, polarization-preserving character of tapered polarization-maintaining fibers (PMFs) is evaluated by OFDR-based distributed birefringence along tapered PMFs. Three tapered-based micro-fiber devices have been used as effective mode selecting components to build narrow-linewidth tunable Erbium-doped fiber ring lasers. The fabrication is easy and at a low cost. 1) a tapered fiber tip forms multimode interference mechanism; 2) a two-taper MZI has been demonstrated by splitting/combining the fundamental mode and higher-order modes through fiber tapers and is tuned by bending one taper waist; 3) a novel tunable fiber Fabry-Perot filter, consisting of a hollow-core photonic bandgap fiber and a micro-fiber, is employed in the reflection mode.

Em áreas que demandam componentes de alto desempenho como a indústria automotiva, aeronáutica e aeroespacial, a otimização do desempenho dinâmico de estruturas é buscada através de diferentes abordagens, como o projeto de materiais específicos à aplicação, ou otimização estrutural topológica. Em particular, o método de otimização estrutural evolucionária bidirecional BESO (Bi-directional Evolutionary Structural Optimization) tem sido utilizado no projeto simultâneo de estruturas hierárquicas, o que significa que o domínio estrutural consiste não somente na estrutura como também na topologia microestrutural dos materiais empregados. O objetivo desse trabalho consiste em aplicar a metodologia BESO na resolução de problemas multiescala bidimensionais visando à maximização da frequência fundamental de estruturas, assim como a minimização de sua resposta quando sujeitas a excitações forçadas numa determinada faixa de frequências. O método da homogeneização é introduzido e aplicado na integração entre as diferentes escalas do problema. Em especial, o modelo de interpolação material é generalizado para o uso de dois materiais no caso de otimização da resposta no domínio da frequência. A metodologia BESO foi aplicada a casos de otimização tomando como domínio estrutural somente a macroescala (projeto estrutural), somente a microescala (projeto material), assim como ambas as escalas concomitantemente (projeto multiescala). Para os casos estudados, a redistribuição de material na macroescala levou a resultados melhores em relação à otimização que modifica a microestrutura. Para a maximização da frequência fundamental, a otimização multiescala obteve os melhores resultados, já para a minimização da resposta em frequência, a otimização somente na macroescala se mostrou mais eficiente. / In areas which demand high performance components, such as automotive, aeronautics and aerospace, the design of application deppendent materials and structural topology optimization are two approaches used in order to optimize structures‟ dynamic behaviour. In particular, the Bi-directional Evolutionary Structural Optimization (BESO) method has been applied to the simultaneous project of hierarchical structures, meaning that the project‟s domain consists not only on the structure on the macroscale, but also on the representative volume element (RVE) associated with the microstructure of the employed materials. The objective of this work is to apply the BESO method in order to solve multiscale bidimensional problems, more specifically, topology optimization problems for fundamental frequency maximization and minimization of the response in the frequency domain under harmonic excitation. The homogenization method is introduced and used to integrate the macro and microscales considered. Furthermore, the material interpolation model in generalized for two material domains in the response minimization problem. The BESO method was applied to optimizations problems where the structural domain was eiher the macrostructure (structural project), microstructure (material project), or both scales simultaneously (multiscale project). In general, material distribution at the macroscale lead to better results in comparison to optimization at the microscale. For fundamental frequency maximization, the multiscale approach obtained better results, while for minimization of the frequency response the results were optimal when the structural domain was restricted to the macrostructure.

In this dissertation, I examine the impact of uncertainty and information processing restrictions on standard economic models. Chapter 1 examines a reevaluation of the excess volatility puzzle in asset prices by assessing the impact of a shift in the agent's focus from minimizing average loss to minimizing maximum loss. Chapters 2 and 3 extend and clarify the newly developing arena of economic models in which the agent's capacity for information processing is systematically limited, as in the recent rational inattention literature. Chapter 1, which represents joint work with Charles Whiteman, studies the consequences changing the present value formula for stock prices. In place of the squared-error-loss minimizing expected present value of future dividends, we use a predictor optimal for the min-max preference relationship appropriate in cases of ambiguity. With ``robust" predictions, the well-known variance bound is reversed in that prices are predicted to be far more volatile than what is observed. We also investigate an intermediate ``partially robust'' case in which the degree of ambiguity is limited, and discover that such an intermediate model cannot be rejected in favor of an unrestricted time series model. Chapter 2 demonstrates the properties and solutions for the more general two-period rational inattention model. We show that the problem is convex, can be solved in seconds, and highlights several important features of information-processing-capacity-constrained models. Additionally, we show the importance of deriving, rather than assuming, the form of the final solution in rational inattention models. Chapter 3 extends the work of Chapter 2 to a finite-horizon dynamic setting by creating a structure in which distributional state and control variables interact under information-processing constraints. Limited information processing capacity is used optimally, and agents have the opportunity to trade processing capacity for higher expected future income. The framework is applied to the canonical life-cycle model of consumption and saving, and an analysis of the impact of preference parameters on optimal attention allocation is conducted. The model produces a distinct hump-shaped profile in expected consumption.

Fiber tapers have attracted much attention and have been successfully employed in various applications, ranging from resonators, filters, interferometers to sensors. This thesis studies the properties of fiber tapers for the purpose of making tapered-based devices and sensors in aerospace related application where small size and light weight are critical. This thesis includes theoretical derivation and experimental verifications of distributed mode coupling in tapered single-mode fibers (SMFs) with high-resolution optical frequency-domain reflectometry (OFDR) technique. The studies are realized with OFDR through phase detection of a Mach-Zehnder interferometer (MZI), which measures local refractive index change relative to the reference arm. The wavelength shifts converted by the phase change give the group index differences between the fundamental mode and higher-order modes of fiber tapers. The energy re-distribution is observed in Rayleigh backscatter amplitude as a function of fiber length with a ~13µm resolution over the entire fiber taper, and group index difference between core and cladding modes is measured with a spatial resolution of ~2cm by using autocorrelation data processing. The thermal and mechanical properties of fiber tapers have also been characterized with OFDR. The cross-correlation wavelength shift is related to the refractive index change of the modes. It is shown that residual stress induced by the tapering process results in the inhomogeneous thermal property, which can be significantly reduced by an annealing treatment. A fiber taper with a waist diameter of ~6µm has a force sensitivity of ~620.83nm/N, ~500 times higher than that of SMF. Furthermore, polarization-preserving character of tapered polarization-maintaining fibers (PMFs) is evaluated by OFDR-based distributed birefringence along tapered PMFs. Three tapered-based micro-fiber devices have been used as effective mode selecting components to build narrow-linewidth tunable Erbium-doped fiber ring lasers. The fabrication is easy and at a low cost. 1) a tapered fiber tip forms multimode interference mechanism; 2) a two-taper MZI has been demonstrated by splitting/combining the fundamental mode and higher-order modes through fiber tapers and is tuned by bending one taper waist; 3) a novel tunable fiber Fabry-Perot filter, consisting of a hollow-core photonic bandgap fiber and a micro-fiber, is employed in the reflection mode.

Evaluation of motion characteristics of ships and offshore structures at the early stage of design as well as during operation at the site is very important. Strip theory based programs and 3D panel method based programs are the most popular tools used in industry for vessel motion analysis. These programs use different variations of the Green’s function or Rankine sources to formulate the boundary element problem which solves the water wave radiation and diffraction problem in the frequency domain or the time domain. This study presents the development of a 3D frequency domain Green’s function method in infinite water depth for predicting hydrodynamic coefficients, wave induced forces and motions. The complete theory and its numerical implementation are discussed in detail. An in house application has been developed to verify the numerical implementation and facilitate further development of the program towards higher order methods, inclusion of forward speed effects, finite depth Green function, hydro elasticity, etc. The results were successfully compared and validated with analytical results where available and the industry standard computer program WAMIT v7.04 for simple structures such as floating hemisphere, cylinder and box barge as well as complex structures such as ship, spar and a tension leg platform.

In this thesis, we demonstrated the effectiveness and the accuracy of the FD-FD method for complex optical waveguide structures such as the micro ring resonator, micro disk resonator, tapered waveguides and waveguides terminated with tilted facets. We are able to achieve the goals by deriving the following modification/extension of the original FD-FD methods. In frequency domain, we can build an accurate frequency-domain modal absorbing boundary condition (ABC) for both the homogeneous region and for the layered background. This allows us to connect the analytical modal solutions with FD solutions and thus reduce the area of the FD domain. In addition, we adopt an effective index averaging method for representing equivalent material for grid cells containing more than one kind of materials. For the TM case, for each grid cell we need to compute effective indices for all four surrounding cells (left, right, up, and down). For the TE case, we need to compute just one effective index within each grid cell. Note that we employ two different averaging schemes for the TE and the TM cases. To solve the huge block tri-diagonal matrix equation (derived from the FD-FD approximation) we modified the Thomas method and we were able to obtain the solutions of linear equations involving more than a hundred thousand variables under a few minutes. We used our method to analyze optical micro-ring waveguides, micro-disk cavities, adiabatic tapered waveguides and waveguides terminated with tilted facets. The simulated results include the reflection coefficients, transmission coefficients and field distribution.

In this work, the circuit implementation of the front-end for Frequency Domain (FD) Sampling Receiver is presented. Shooting for two different applications, two transconductance amplifiers are designed. A high linear transconductance amplifier with 25 dBm IIP3 is proposed to form the high resolution and high sampling rate FD receiver. The whole system achieves an overall sampling rate of 2 Gs/s and resolution of 10 bits. Another low noise transconductance amplifier exploiting noise cancelling is designed to build up the FD wireless communication receiver, which is an excellent candidate for Software Define Radio (SDR) and Cognitve Radio (CR). The proposed noise cancelling scheme can suppress both thermal noise and flicker noise at the frontend. The system Noise Figure (NF) is improved by 3.28 dB. The two transconductance amplifiers are simulated and fabricated with TI 45nm CMOS technology.

The continuous phase modulation (CPM) has a constant envelope and compact output power spectrum that makes it a promising underlying technology for power and spectrum efficient broadband wireless communications. However, high implementation complexity (especially the complexity of the receiver) required to deal with the phase memory and inter-symbol interference has impeded its adoption for broadband wireless communications, and only a few simple CPM modulation schemes have mainly been used, e.g. binary MSK and GMSK. Thus, research on efficient CPM transceivers to reduce the computational and hardware complexity is important. The major contribution of this dissertation is the development of novel frequency domain processing techniques and transceiver strategies to improve power and spectral efficiency, and reduce the complexity of CPM modulation schemes. First, this dissertation presents simplified frequency domain receiver structures and decoding schemes in the frequency domain for binary and M-ary CPM block transmission. The frequency domain receivers utilize parallel and serial structures with frequency domain processing which considerably reduces hardware and computational complexity compared to conventional time-domain processing. In addition, the decoding schemes in the frequency domain eliminate the controlled phase memory through frequency domain phase equalization instead of maximum-likelihood sequential decoders, e.g. Viterbi decoders. Second, frequency domain channel estimation schemes for CPM block transmission are presented, which adopt superimposed training signals to achieve bandwidth and power efficiency while reducing the complexity. In these schemes, the proposed frequency domain channel estimation uses the superimposed training signals as a reference signal to reduce the throughput loss caused by conventionally multiplexed training signals. Superimposed training signal design is presented, and the trade-off between bandwidth efficiency and power efficiency is also analyzed. Third, block transmission schemes and frequency domain equalization methods for CPM are proposed, which consider linear processing instead of conventional decomposition-based processing. The schemes of frequency domain linear processing avoid the complexity overhead (both in computation and hardware) of conventional orthogonal- or Laurent decomposed-based equalizers. Finally, this dissertation extends CPM block transmission and frequency domain equalization to phase-coded (time-varying modulation index) CPM, which shows better error performance and bandwidth efficiency than fixed modulation index CPM's. / text