In the thesis, we present the structures of interacting multiple model (IMM) algorithm-based filter design for real-time motion orientation estimation and tracking by using inertial sensor measurements in three-dimensional space. The major sensor such as gyroscope, though has high-sensitivity characteristics, suffers from bias build-up and error drift over time. The complementary sensors such as accelerometer and magnetometer, on the other hand, have low sensitivity, but do not suffer from bias problems. By using individual inertial and magnetic sensors, measurements of multiple modes can be interactively computed. The IMM based designs show the advantages of weighting individual sensors in different motion states. We propose a signal processing architecture based on the IMM algorithm. It is composed of three parallel Kalman filters (KFs), each deals with measured signals from accelerometer, magnetometer and gyroscope, respectively. The accelerometer cannot effectively sense the rotation around the vertical axis; while the magnetometer can only sense the rotation around vertical axis. Therefore, estimation accuracy with the parallel filtering arrangement of the IMM algorithm-based structure may be affected. A scheme using the residual signal, which is computed in the IMM, provides the information of gyroscope-based KF to the other two filters for feasible calculation of update weights. Related research also usually combined the information of major and complementary sensors in estimator designs. In the literature, existing ¡§Triad¡¨ methods with quaternion-based extended Kalman filter (EKF), process the measurements from major and complementary sensors. To compensate the functions, we propose to use a gyroscope-based EKF and a Triad EKF in forming a parallel multiple model-based structure. The analysis and performance evaluation shows advantages and disadvantages of using EKFs and KFs in IMM-based filtering approachs. Simulation results validate the proposed estimator design concept, and show that the scheme is capable of reducing the overall estimation errors by flexible computation of model weights.

In this thesis, the blind phase estimator algorithms are studied, including Histogram Algorithm (HA), Modified Histogram Algorithm (MHA), Maximum Likelihood (ML), 4th power-law and modified 4th power-law to compare their performance under quadrature amplitude modulation and additive white gaussian noise(AWGN) channel. Owing to the development of wireless communication, I used the fading channel of wireless communication and studied the performance of all algorithms again. After differential encoding and quadrature amplitude modulation, signals enter additive white gaussian noise(AWGN), constant phase offset and fading channel. At receiver, I use the above estimator to find the phase offset. If there is not line of sight under fading channel, all estimators performance are not good from simulation results, if there is a strong line of sight, all estimators performance are good and approximate to the AWGN channel.

Blind estimation of unknown channel parameters and data symbol detection represent major open problems in non-cooperative communication systems such as automatic modulation classification (AMC). This thesis focuses on estimating the symbol rate and detecting the data symbols. A blind oversampling-based signal detector under the circumstance of unknown symbol period is proposed. The thesis consists of two parts: a symbol rate estimator and a symbol detector. First, the symbol rate is estimated using the EM algorithm. In the EM algorithm, it is difficult to obtain the closed form of the log-likelihood function and the density function. Therefore, both functions are approximated by using the Particle Filter (PF) technique. In addition, the symbol rate estimator based on cyclic correlation is proposed as an initialization estimator since the EM algorithm requires initial estimates. To take advantage of the cyclostationary property of the received signal, there is a requirement that the sampling period should be at least four times less than the symbol period on the receiver side. Second, the blind data symbol detector based on the PF algorithm is designed. Since the signal is oversampled at the receiver side, a delayed multi-sampling PF detector is proposed to manage inter-symbol interference, which is caused by over- sampling, and to improve the demodulation performance of the data symbols. In the PF algorithm, the hybrid importance function is used to generate both data samples and channel model coe±cients, and the Mixture Kalman Filter (MKF) algorithm is used to marginalize out the fading channel coe±cients. At the end, two resampling schemes are adopted.

Given a certain transmission frequency, Shannon spatial sampling limit de¯nes an upper bound for the antenna element spacing. Beyond this bound, the exceeded ambiguity avoids correct estimation of the signal parameters (i.e., array manifold crossing). This spacing limit is inversely proportional to the frequency of transmis- sion. Therefore, to meet a wider spectral support, the element spacing should be decreased. However, practical implementations of closely spaced elements result in a detrimental increase in electromagnetic mutual couplings among the sensors. Further- more, decreasing the spacing reduces the array angle resolution. In this dissertation, the problem of Direction of Arrival (DOA) estimation of broadband sources is ad- dressed when the element spacing of a Uniform Array Antenna (ULA) is inordinate. It is illustrated that one can resolve the aliasing ambiguity by utilizing the frequency diversity of the broadband sources. An algorithm, based on Maximum Likelihood Estimator (MLE), is proposed to estimate the transmitted data signal and the DOA of each source. In the sequel, a subspace-based algorithm is developed and the prob- lem of order estimation is discussed. The adopted signaling framework assumes a subband hopping transmission in order to resolve the problem of source associations and system identi¯cation. The proposed algorithms relax the stringent maximum element-spacing constraint of the arrays pertinent to the upper-bound of frequency transmission and suggest that, under some mild constraints, the element spacing can be conveniently increased. An approximate expression for the estimation error has also been developed to gauge the behavior of the proposed algorithms. Through con- ¯rmatory simulation, it is shown that the performance gain of the proposed setup is potentially signi¯cant, speci¯cally when the transmitters are closely spaced and under low Signal to Noise Ratio (SNR), which makes it applicable to license-free communication.

The angular resolution of an imaging system is sharply bounded by the diffraction limit, a fundamental property of electromagnetic radiation propagation. In order to increase resolution and see finer details of remote objects, the sizes of telescopes and cameras must be increased. As the size of the optics increase, practical problems and costs increase rapidly, making sparse aperture systems attractive for some cases. The method of Intensity Correlation Imaging (ICI) provides an alternative method of achieving high angular resolution that allows a system to be built with less stringent precision requirements, trading the mechanical complexity of a typical sparse aperture for increased computational requirements. Development of ICI has stagnated in the past due to the inadequacies of computational capabilities, but the continued development of computer technologies now allow us to approach the image reconstruction process in a new, more e ffctive manner. This thesis uses estimation methodology and the concept of transverse phase diversity to explore the modern bounds on the uses of ICI. Considering astronomical observations, the work moves beyond the traditional, single-parameter uses of ICI, and studies systems with many parameters and complex interactions. It is shown that ICI could allow significant new understanding of complex multi-star systems. Also considered are exoplanet and star-spot measurements; these are less promising due to noise considerations. Looking at the Earth imaging problem, we find significant challenges, particularly related to pointing requirements and the need for a large field-of-view. However, applying transverse phase diversity (TPD) measurements and a least-squares estimation methodology solves many of these problems and re-opens the possibility of applying ICI to the Earth-imaging problem. The thesis presents the TPD concept, demonstrates a sample design that takes advantage of the new development, and implements reconstruction techniques. While computational challenges remain, the concept is shown to be viable. Ultimately the work presented demonstrates that modern developments greatly enhance the potential of ICI. However, challenges remain, particularly those related to noise levels.

Facility managers use various cost models and techniques to estimate the cost of renovating a building and to secure the required funds needed for building renovation. A literature search indicates that these techniques offer both advantages and disadvantages that need to be studied and analyzed. Descriptive statistical methods and qualitative analysis are employed to identify and compare techniques used by facility managers to calculate the expected renovation costs of a building. The cost models presently used to predict the cost and accumulate the budget required for renovation of a building were determined through interviews with ten Texas-based university facilities managers. The data and information gathered were analyzed and compared. Analysis of results suggests that traditional methods like Floor Area Method (FAM) is the most accurate, less time consuming, easy to use as well as convenient for data collection. Case-Based Reasoning (CBR), though not as widely used as FAM, is known to facilities managers. This is due to the fact that, if a new type of project needs to be renovated, and the data for a similar project is not available with the facilities manager, a completely new database needs to be created. This issue can be resolved by creating a common forum where data for all types of project could be made available for the facilities managers. Methods such as regression analysis and neural networks are known to give more accurate results. However, of the ten interviewees, only one was aware of these new models but did not use them as they would be helpful for very large projects and they would need expertise. Thus such models should be simplified to not only give accurate results in less time but also be easy to use. These results may allow us to discuss changes needed within the various cost models.

Tall building frames that respond to large seismic ground motions tend to have significant spatial variability of damage over their height, often with a concentration of that damage in the lower stories. In spite of this spatial variability of damage, existing damage and loss models tend to focus on taking the maximum story drift and then assuming the same drift applies over the entire height, damage is then calculated for the building—clearly a conservative approach. A new loss analysis approach is thus recommended that incorporates the effects of spatial distribution of earthquake induced damage to frame buildings. Moreover, the approach aims to discriminate between required repair and replacement damages. Suites of earthquakes and incremental dynamic analysis along with the commercial software SAP2000 are used to establish demands from which story damage and financial losses are computed directly and aggregated for the entire structure. Rigorous and simplified methods are developed that account for spatial distribution of different damage levels arising from individual story drifts.

Digital image data requires large compression ratio in applications like internet, communication and audio-visual environment. In this thesis, we realize the MPEG-4 codec standard on the ARM9-based platform and improve the execution performance by efficient implementations of the core operations such as Motion Estimation and DCT. In the assembly codes obtained by directly compiling the C codes, there exists a lot of redundant checking which causes a large amount of execution time waste. We rewrite some of the compiled assembly codes to improve the execution efficiency using a variety of techniques such as loop-unrolling and data-type optimization. We also analyze the experimental results using several benchmark video sequences with different modes.

In this thesis, we propose a modified MT-CDMA system, which can improve channel estimation accuracy by using transmit diversity of pilot signals. We not only expound the principles and structures of the system we proposed, but analyze its performance in slow Rayleigh fading channel environment. According to different ways to assign data symbols of transmitted signals, we have two different bit error rate results. At the same transmit power, the simulation results show that when we combine comb-type pilot signals of two parallel channels to estimate channel gains, we can recover the drawbacks of comb-type pilot arrangement, which can not perform well in frequency selective fading channel. Finally, the numerical results will be also shown.

A complex array for constructing perfect sequences is presented in this paper. The row sequences and their discrete Fourier transform form two sets of perfect sequences. The column sequences are orthogonal to each other for any cyclic shift. In addition, any combination of the column sequences with complex weighting coefficients of equal amplitude is also a perfect sequence. In addition, a superimposed training scheme is also proposed for channel estimation in OFDM systems. The perfect sequence is adopted since it has a constant magnitude in both the time domain and the frequency domain. Although the derived channel estimator has a slightly worse performance since the unknown data contributes extra noise, the effective data throughput is substantially increased. In addition, the proposed scheme is shown to have a much better peak-to-average power ratio (PAPR) because the added perfect sequence has a constant magnitude in the time domain.