Thesis (M.Eng.Stud.)--University of Wollongong, 2008. / Typescript. Includes bibliographical references: leaf 84-95.

Project (M.Eng.) - Simon Fraser University, 2004. / Theses (School of Engineering Science) / Simon Fraser University. Also available on the World Wide Web.

Thesis (Ph. D.)--Electrical and Computer Engineering, Georgia Institute of Technology, 2006. / Dr. Alfred Andrew, Committee Member ; Dr. Ye (Geofferey) Li, Committee Member ; Dr. Nikil S. Jayant, Committee Member ; Dr. Gordon L. Stuber, Committee Chair ; Dr. Douglas B. Williams, Committee Member Vita. Includes bibliographical references.

This thesis presents algorithms for performing the next two reduction steps, namely orthogonalization and extraction. More specifically, this thesis addresses the following research question: What are proper methods of orthogonalizing spectral images in preparation for extraction?

Abstract not available

The general objective of this proposed research is to design and develop signal acquisition and tracking algorithms for multiple input multiple output orthogonal frequency division multiplexing (MIMO-OFDM) systems for fixed wireless access applications. The algorithms are specifically targeted for systems that work in time division multiple access and frequency division multiple access frame modes. In our research, we first develop a comprehensive system model for a MIMO-OFDM system under the influence of the radio frequency (RF) oscillator frequency offset, sampling frequency (SF) offset, RF oscillator phase noise, frequency selective channel impairments and finally the additive white Gaussian noise. We then develop the acquisition and tracking algorithms to estimate and track all these parameters. The acquisition and tracking algorithms are assisted by a preamble consisting of one or more training sequences and pilot symbol matrices. Along with the signal acquisition and tracking algorithms, we also consider design of the MIMO-OFDM preamble and pilot signals that enable the suggested algorithms to work efficiently. Signal acquisition as defined in our research consists of time and RF synchronization, SF offset estimation and correction, phase noise estimation and correction and finally channel estimation. Signal tracking consists of RF, SF, phase noise and channel tracking. Time synchronization, RF oscillator frequency offset, SF oscillator frequency offset, phase noise and channel estimation and tracking are all research topics by themselves. A large number of studies have addressed these issues, but usually individually and for single-input single-output (SISO) OFDM systems. In the proposed research we present a complete suite of signal acquisition and tracking algorithms for MIMO-OFDM systems along with Cramr-Rao bounds for the SISO-OFDM case. In addition, we also derive the Maximum Likelihood (ML) estimates of the parameters for the SISO-OFDM case. Our proposed research is unique from the existing literature in that it presents a complete receiver implementation for MIMO-OFDM systems and accounts for the cumulative effects of all possible acquisition and tracking errors on the bit error rate (BER) performance. The suggested algorithms and the pilot/training schemes may be applied to any MIMO OFDM system and are independent of the space-time coding techniques that are employed.

Indiana University-Purdue University Indianapolis (IUPUI) / The Convolutional Neural Network (CNN) is a mathematical model designed to distill input information into a more useful representation. This distillation process removes information over time through a series of dimensionality reductions, which ultimately, grant the model the ability to resist noise, and generalize effectively. However, CNNs often contain elements that are ineffective at contributing towards useful representations. This Thesis aims at providing a remedy for this problem by introducing Absolute Cosine Value Regularization (ACVR). This is a regularization technique hypothesized to increase the representational power of CNNs by using a Gradient Descent Orthogonalization algorithm to force the vectors that constitute their filters at any given convolutional layer to occupy unique positions in in their respective spaces. This method should in theory, lead to a more effective balance between information loss and representational power, ultimately, increasing network performance. The following Thesis proposes and examines the mathematics and intuition behind ACVR, and goes on to propose Dynamic-ACVR (D-ACVR). This Thesis also proposes and examines the effects of ACVR on the filters of a low-dimensional CNN, as well as the effects of ACVR and D-ACVR on traditional Convolutional filters in VGG-19. Finally, this Thesis proposes and examines regularization of the Pointwise filters in MobileNetv1.

Wireless devices in wireless networks are powered typically by small batteries that are not replaceable nor recharged in a convenient way. To prolong the operating lifetime of networks, energy efficiency is indicated as a critical issue and energy-efficient resource allocation designs have been extensively developed. We investigated energy-efficient schemes that prolong network operating lifetime in wireless sensor networks and in wireless relay networks. In Chapter 2, the energy-efficient resource allocation that minimizes a general cost function of average user powers for small- or medium-scale wireless sensor networks, where the simple time-division multiple-access (TDMA) is adopted as the multiple access scheme. A class of Ç-fair cost-functions is derived to balance the tradeoff between efficiency and fairness in energy-efficient designs. Based on such cost functions, optimal channel-adaptive resource allocation schemes are developed for both single-hop and multi-hop TDMA sensor networks. In Chapter 3, optimal power control methods to balance the tradeoff between energy efficiency and fairness for wireless cooperative networks are developed. It is important to maximize power efficiency by minimizing power consumption for a given quality of service, such as the data rate; it is also equally important to evenly or fairly distribute power consumption to all nodes to maximize the network life. The optimal power control policy proposed is derived in a quasi-closed form by solving a convex optimization problem with a properly chosen cost-function. To further optimize a wireless relay network performance, an orthogonal frequency division multiplexing (OFDM) based multi-user wireless relay network is considered in Chapter 4. / In the OFDM approach, each subcarrier is dynamically assigned to a source- destination link, and several relays assist communication between pairs of source-destination over their assigned subcarriers. Using a class of Ç-fair cost-functions to balance the tradeoff between energy efficiency and fairness, jointly with optimal subcarrier and power allocation schemes at the relays. Relevant algorithms are derived in quasi-closed form. Lastly, the proposed energy-efficient schemes are summarized and future work is discussed in Chapter 5. / by Di Wang. / Thesis (Ph.D.)--Florida Atlantic University, 2011. / Includes bibliography. / Electronic reproduction. Boca Raton, Fla., 2011. Mode of access: World Wide Web.

The purpose of this thesis is to implement algorithms of matrix eigendecomposition and pseudoinverse computation on a Field Programmable Gate Array (FPGA) platform. Firstly, there are described matrix decomposition methods that are broadly used in mentioned algorithms. Next section is focused on the basic theory and methods of computation eigenvalues and eigenvectors as well as matrix pseudoinverse. Several examples of implementation using Matlab are attached. The Vivado High-Level Synthesis tools and libraries were used for final implementation. After the brief introduction into the FPGA fundamentals the thesis continues with a description of implemented blocks. The results of each variant were compared in terms of timing and FPGA utilization. The selected block has been validated on the development board and its arithmetic precision was analyzed.

The Convolutional Neural Network (CNN) is a mathematical model designed to distill input information into a more useful representation. This distillation process removes information over time through a series of dimensionality reductions, which ultimately, grant the model the ability to resist noise, and generalize effectively. However, CNNs often contain elements that are ineffective at contributing towards useful representations. This Thesis aims at providing a remedy for this problem by introducing Absolute Cosine Value Regularization (ACVR). This is a regularization technique hypothesized to increase the representational power of CNNs by using a Gradient Descent Orthogonalization algorithm to force the vectors that constitute their filters at any given convolutional layer to occupy unique positions in Rⁿ. This method should in theory, lead to a more effective balance between information loss and representational power, ultimately, increasing network performance. The following Thesis proposes and examines the mathematics and intuition behind ACVR, and goes on to propose Dynamic-ACVR (D-ACVR). This Thesis also proposes and examines the effects of ACVR on the filters of a low-dimensional CNN, as well as the effects of ACVR and D-ACVR on traditional Convolutional filters in VGG-19. Finally, this Thesis proposes and examines regularization of the Pointwise filters in MobileNetv1.