Search

With the growth of information dissemination over digital communication networks, much research has been devoted to compressing digital image information for efficient transmission. The ability to adjust the desired resolution of an image as the available bandwidth on the network changes allows the user to control the flow of data according to the resources available. In this thesis we integrate multiresolution image compression methods with image recognition techniques to assist in automatic image recognition at several levels of resolution. Features of grayscale and binary images of text characters and aircraft line drawings are described using wavelet transform coefficients, wavelet transform subband energy, and Fourier transform coefficients. Transmission of these features over a digital communication link is simulated, and multiresolution recognition performance in the presence of channel noise is presented. http://archive.org/details/multiresolutioni1094532113 NA U.S. Navy (U.S.N.) author Approved for public release; distribution is unlimited.

This thesis presents the bandpass inverse fast Fourier transform (IFFT) filter bank and the multirate digital filter bank techniques to synthesize test point waveforms from constituent waveforms recorded by two instruments as part of an aircraft electromagnetic hardness evaluation test. The component waveforms are recorded by two separate measurement systems (High Powered Pulse Waveform (HPW) in the time domain and Continuous Sweep Waveform (CSW) in the frequency domain) under two different aircraft orientations (parallel and perpendicular). Data from two orientations are combined using the sinusoidal modeling algorithm (SMA). The tree structured filter bank with power symmetric overlap method and the bandpass IFFT with spectral concatenation method are developed to further combine these waveforms with an overlapping frequency spectrum to produce the corresponding synthesized test point waveform http://archive.org/details/widebandsignalna1094532168 NA Republic of Korea Army author Approved for public release; distribution is unlimited.

This thesis develops an algorithm to fuse redundant observations due to multiple sensor coverage of a vessel. Fuzzy membership functions are used as a measure of correlation, and a fuzzy associative system determines which observations represent the same vessel. The result is a computationally efficient algorithm. The output of the system is a unique set of vessels identified by unique platform identifiers. Results of tests based on computer simulation of overlapping radar coverage show that the fusion algorithm correctly correlates and fuses the sensor observations. That the VTS system is a subset of the Joint Maritime Command Information System (JMCIS) and ultimately the Global Command and Control Software (GCCS) system makes this algorithm pertinent not only to the US Coast Guard, but also to the US Navy, DOD and other agencies such as the Canadian Navy that use this software. http://archive.org/details/multileveldatass1094531314 NA NA Canadian Armed Forces author.

With the advances in high speed, programmable digital signal processing (DSP) chips, modern communications links are using a combination of DSP techniques and digital communications methods to realize faster, reconfigurable, and modular systems. This thesis details the software implementation of a modern digital communication system combining various DSP functions, channel Forward Error Correcting (FEC) algorithms, and digital modulation methods. The digital modulation schemes considered here include both baseband and Quadrature Phase Shift Keying (QPSK) techniques. The proposed communication system will serve as a practical tool useful for simulating the transmission of any digital data. The various modules of the system include source encoders/decoders, data compression functions, channel encoders/decoders, and modulators/demodulators. Implementation consists of coding the various link functions in C and integrating them as a complete system. The results show the viability of a QPSK modulated digital communications link and point the direction of future research towards software radio. http://archive.org/details/implementationof1094531312 NA NA U.S. Navy (U.S.N.) author.

Autoregressive analysis is used in modern signal processing applications for modeling and estimation of random signals. High speed digital signal processors with advanced architecture and special digital signal processing instructions, mostly compiled in C language, can be used in these applications to achieve realtime performance. A commercially available digital signal processor has been used in this work to estimate the AR parameters and power spectral density from the given input data by using the Levinson, Burg and Schur algorithms. This work produced a library file that contains the object files of the AR parameter estimation algorithms. The time required in terms of the cycle counts to execute each algorithm is listed for different data lengths and model orders. http://archive.org/details/arparameterestim1094531332 NA NA Turkish Navy author.

The advances in high speed computer networks and digital communication techniques have enabled the rapid and extensive dissemination of information throughout the modern defense infrastructure. One of the challenges in networking today is real-time dissemination of information. This thesis proposes a solution for a specific aspect of this challenge, namely, the transmission of real-time compressed video data across the Common Data Link (CDL), a high-speed military data link designed to operate in high error environments. Current research is primarily focused on the transmission of real- time data across low-error links. This thesis proposes, simulates, and analyzes a mechanism which guarantees that (1) delay bounds are met for real-time flows despite network overload and (2) a minimum acceptable image quality is maintained despite the presence of highly correlated errors. These highly correlated errors are characteristic of the type of electromagnetic jamming likely to be encountered by the CDL. This mechanism consists of four fundamental requirements: (1) a hierarchical image compression scheme, (2) rate control at the source, (3) bandwidth allocation within all encountered network nodes, and (4) dynamic forward error correction. The proposed solution is modeled in the OPNET simulation environment, and the validity and feasibility of the mechanism are verified. In addition, the simulation is interfaced with a compression/decompression algorithm running in MATLAB to enable the subjective analysis of actual images before and after transmission in various jamming scenarios. The results demonstrate the effectiveness of the proposed solution in meeting delay guarantees and maintaining image quality. http://archive.org/details/realtimecompress109457522 U.S. Navy (U.S.N.) author.

In this thesis a scheme is proposed for still image compression for realtime transmission over a packet switched network. A layered image compression scheme based on the quad tree decomposition technique (QTD) is developed. The image is decomposed into three layers, where layers are added to a base image to incrementally improve the quality of the reconstructed image. The performance of the QTD is enhanced by adding a DCT algorithm with adaptive bit allocation and a region of interest methodology. The quad tree is then extended to a nona tree whereby pixel blocks of size 3 x 3 are processed instead of 2 x 2. Results are presented to provide a visual subjective perception of the proposed scheme. Depending on the number of layers transmitted, the compression gains range from 60:1 to 16:1. It is shown that the proposed scheme exhibits graceful degradation of the reconstructed image and some error tolerant capabilities. http://archive.org/details/imagecompression109457538 Singapore Ministry of Defense author (civilian).

This thesis explores the use of neural networks to perform multisensor data fusion for Vessel Traffic Services (VTS). It begins with a detailed study of the VTS system in order to identify the type of input data and other system features that are suitable for fusion. This is followed by a brief study of the various neural networks to evaluate their suitability for data fusion applications. The Kohonen's self-organizing feature map (SOFM) was identified as the most suitable neural network that can be used for data fusion, but it has some limitations that make it unsuitable for solving the VTS data fusion problem. A neural network data fusion model was proposed that consists of a modified SOFM and a double fusion resolver to solve the problem of double fusion in VTS. The proposed model is simulated in software and tested with measured input data supplied by the U.S. Coast Guard. Results of fusion tests indicate that the proposed fusion system performs well; thus, the proposed neural network fusion model has potential for implementation in the VTS system. http://archive.org/details/aneuralnetworkpp1094531577 NA NA Republic of Singapore author (civilian).

This thesis develops an algorithm to fuse redundant radar observations caused by multiple radar coverage of a vessel. Additionally, the algorithm automates track history and improves position accuracy by incorporating differential Global Positioning System (GPS) reports. The algorithm uses fuzzy membership functions as a measure of correlation and a fuzzy associative system to determine which observations represent the same vessel. The use of the fuzzy associative system produces a computationally efficient algorithm. A track correlation technique is used to automate the track history thereby reducing operator workload. Results of tests based on computer simulation show that the fusion algorithm correctly correlates and fuses the radar observations. (AN) http://archive.org/details/datfusionlgorith1094535079 NA NA U.S. Navy (U.S.N.) author.

Fundamental results developed by Wiener in the 1950's are combined with new work in the area of higher-order statistics to develop and explore a general model for nonlinear stochastic processes. The Wiener model is developed for discrete nonlin- ear systems and its orthogonality properties are analyzed to characterize its output statistics. An efficient structured procedure for computing the fc th -order statistics of the model output is formulated in both the time and frequency domains. Explicit formulas that exploit the structure of the Wiener model are given for computing the cumulants and polyspectra. A necessary condition for a discrete random process to be representable by the Wiener model is discussed. A computationally efficient pro- cedure is given for matching the model output cumulants to estimated cumulants for a given process by minimizing the squared magnitude of the error. Examples of applying this procedure to given sets of data are presented. http://archive.org/details/analysisofnongau1094542815 Egyptian Army author Approved for public release; distribution is unlimited.

This thesis provides a comprehensive study of data compression standards used by both the military and commercial industry. Text, still imagery, video and audio compression are discussed. Several of the military standards described appear to be lacking in both performance and technology when compared to the international standards available. International standards could support the objectives as well as the functionality of the Global Command and Control System. The international standards could also provide better long term interoperability and information exchange capabilities than the currently available military standards. http://archive.org/details/datcompressionst1094530544 NA NA U.S. Navy (U.S.N.) author.

An empirical analysis on the applicability of multigrid techniques to system modeling using system identification techniques is presented. Multigrid with the Toeplitz approximation algorithm is used to model an infinite impulse response (IIR) system with a finite impulse response (FIR) model. Matrix analysis experiments are conducted on the Toeplitz iteration matrix. A comparison of alternative multigrid operators for solving the system modeling problem is presented. The extension of multigrid techniques to system modeling of odd order filters is explored. Computer simulations are developed for analyzing the effectiveness of multigrid rechniques in system modeling. http://archive.org/details/analysisofmultig1094543037 Naval Postgraduate School author Approved for public release; distribution is unlimited.

High range resolution can be obtained using a number of methods including ultra-wideband radar, intrapulse pulse compression, and super- resolution techniques. This thesis investigates the achievement of high radar range resolution by means of the step frequency waveform. The key advantage of the step frequency approach compared to other methods is the wide total bandwidth resulting in high range resolution while still maintaining a narrow instantaneous bandwidth which eases A/D sampling requirements. The effects of various waveform parameters on the high resolution range profile are determined for both noise and clutter limited environments. A novel technique is developed which enhances the resolution of moving targets in clutter and estimates the target velocity. This method consists of clutter cancellation followed by successive velocity compensations. Finally, design methods are developed to optimize system performance in a low PRF mode for both noise and clutter limited environments http://archive.org/details/highradarresolut1094528262 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

This thesis describes a subsystem developed to provide for the transfer of digital audio signals from a SUN SPARCstation 10 workstation to a digital audio tape (DAT) and vice versa. The new system expands the audio recording/reproduction options available in the laboratory by integrating an analog tape deck and a digital tape deck with the SUN workstation. The desired connection enables working with a larger audio bandwidth to achieve better audio performance and resolution in comparison to the present workstation audio capabilities. Performance measurements of the audio signal-to-noise ratio have shown an improvement of about 45 dB in the audio reproduction capability and about 35 dB in the audio recording capability. Total harmonic distortion for the new system is below the limit of the measuring instrumentation (less than 0.1%). http://archive.org/details/hifidigitaludiot1094530908 Lieutenant Commander, Israeli Navy Approved for public release; distribution is unlimited.

In this thesis, a hybrid lossless compression model is tested which employs a combination of both a lossy compression method and one or more lossless image compression methods to produce an overall lossless image compression. The hybrid model decomposes the original image into a browse and a residual image. The hybrid model is tested and evaluated using various combinations of lossy and lossless image compression methods. The lossy compression method used in the model is JPEG (Joint Photographic Experts (Group) . The lossless compression methods used are Huffman, Arithmetic, LZW, lossless JPEG. and Diagonal coding. The compression results achieved using the hybrid compression model are compared to the compression achieved using the corresponding direct lossless compression. Additionally, the hybrid model is evaluated as to the advantages that the decomposition of the image into browse and residual images provide to the user. http://archive.org/details/asimplelowoverhe1094527264 Lieutenant, United States Navy

A multiple-input, multiple-output (MIMO) recursive least squares (RLS) algorithm is developed to shape and control the Loran-C RF pulse of the AN/FPN-44A tube type transmitter. The control algorithm is incorporated into a transmitter simulation program, where it seeks to produce an optimal transmitter drive waveform (TDW). An optimal TDW produces a near ideal RF pulse. The control algorithm uses a MIMO reference model of the transmitter; parameters of the model are obtained using recursive least squares multichannel time series techniques. The MIMO reference model has the ability to adapt to the non-LTI characteristics of the simulated transmitter. The MMO RLS control algorithm is implemented in both an ideal and a realistic noisy environment. In the ideal environment, when representing the RF pulse with parameters of its half-cycle peak amplitudes and zero-crossings, the MIMO RLS controller is able to shape the RF pulse and control its zero-crossings. Quantization and system noise in the non-ideal environment results in performance deterioration of the control algorithm. The performance of the MIMO RLS algorithm is compared against another method of control, the steepest descent algorithm. http://archive.org/details/mimorecursivelea1094540014 Lieutenant, United States Coast Guard Approved for public release; distribution is unlimited.

This thesis proposes a new analysis/synthesis procedure for speech and image compression. The algorithm applies the discrete wavelet transform to subject data in order to obtain a set of multiresolution wavelet coefficients. The wavelet coefficients are then encoded by using the generalized Lloyd algorithm. The statistical properties of the wavelet coefficients are utilized to determine the number of resolution levels as well as the codebook size at each resolution level. Coding results show that the new procedure provides a significant improvement in the quality of the reproduced data. The tested data includes speech, image, and transient signals. http://archive.org/details/datcompressionby1094539786 Lieutenant, J.G., Turkish Navy Approved for public release; distribution is unlimited.

Automatic pulse shape control is simulated for the N/FPN-42 and AN/FPN-44A tube type transmitters. A linear, time invariant (LTI) pole-zero model is developed for each transmitter at a typical operating point using the least squares modified Yule-Walker method and Shank's method. LTI models for a range of operating points are catenated to represent observed nonlinear behavior, and observed time variations are added. After these combined models are tested, a linear controller based on the method of steepest descent is implemented. These models, the control algorithm and transmitter system details such as power supply droop, dual rating and noise are then incorporated into a MATLAB simulation program. In a variety of realistic tests the control algorithm successfully shaped the Loran-C pulse, except that zero-crossing times were not always in tolerance and the algorithm showed a sensitivity to noise. The algorithm controlled Envelope-to-Cycle Difference, produced an entire Phase Code Interval of pulses while compensating for droop and phase code bounce, and produced a near- optimal transmitter drive waveform for the transmitter/antenna system using the dummy load. http://archive.org/details/automaticpulsesh1094538503 Lieutenant, United States Coast Guard Approved for public release; distribution is unlimited.

This thesis is concerned with the use of simulation in the teaching of radar signal processing (RSP). The aspects of RSP to be investigated and simulated are the development of the DFT as a filter bank for radar applications, filter sidelobe reduction by the use of weighted DFT Doppler filter banks and the generation of pulse compression coding schemes under Doppler conditions. Simulation programs have been written in MATLAB for the above processes. these simulations are cost effective, convenient to use and easy to reproduce since they are run on personal computers that are readily available to students. This provides a learning environment which cannot be achieved by traditional methods alone and compliments classroom teaching. In particular it aids the student by removing a major computational burden which allows the student to explore thee non-trivial real world problems which he could not do before. http://archive.org/details/developmentofrad1094523637 Major, Canadian Army Approved for public release; distribution is unlimited.

This thesis develops an iterative algorithm for the design of ARMA models of signals in the time domain. The algorithm is based on optimization techniques, particularly a gradient technique known as the restricted step method is used. The new algorithm is called the iterative Prony method, and the results obtained using this new method are compared to those obtained using the iterative prefiltering algorithm. The thesis shows that the performance of the iterative Prony method is in most of the cases comparable or superior to that of the iterative prefiltering algorithm. http://archive.org/details/armamodelingofsi1094523820 Lieutenant Commander, Colombian Navy Approved for public release; distribution is unlimited.

The aim of this thesis work is to explore the use of fuzzy systems in a speech coding and classification application. A mixed excitation LPC based speech coder is developed. The excitation classifier for the speech coder is then implemented using a fuzzy system. The fuzzy logic based classifier determines the type of excitation to be used in constructing the synthetic speech. The results of various implementations of this speech coder are presented for comparison. This work demonstrates that a fuzzy system can be developed and implemented for a classification problem. http://archive.org/details/amixedexcitation1094523960 Lieutenant, United States Coast Guard Approved for public release; distribution is unlimited.

This thesis examines the combination of three algorithms: Graph Theoretic Tracker (GTT), Hough Transform, and Heuristic Search to enhance the detection of spectral tracks of underwater targets in LOFARGRAMS. Previous studies examined these algorithms separately. Here, GTT is used as a pre-processor of the LOFARGRAM display data to locat optimum paths of signals through noise. The line tonals in the output image from GTT are then manipulated by the Hough Transform into clusters of points in parameter space. A Heuristic Search sorting technique is employed to determine cluster centers. These cluster centers are then reconstructed back into line tonals using the inverse Hough Transform formula. The results of this thesis show improvements by taking the Hough Transform of the original LOFARGRAM masked by the output of GTT. The effect of background noise is offset by the accumulation in the parameter space. Subsequently, the recovery of desired tonals is improved. http://archive.org/details/acombinatorialpp1094526922 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

Discrete orthogonal functions are used in adaptive system identification algorithms. Adaptive filters are realized by forming linear combinations of discrete Legendre, Laguerre, and Jacobi polynomials, and backward prediction-error polynomials from a lattice structure. The adaptive filter weights are updated using the LMS algorithm. FIR and IIR bandpass filters are modeled using the adaptive filters, and performance comparisons are made. http://archive.org/details/adaptivefilterde1094526848 Lieutenant, United States Navy

This thesis describes the design of a fully digital Binary Phase Shift Keyed receiver considered for use in a telemetry system in the range testing of torpedos. The method employed uses a subspace technique called Estimation of Signal Parameters via Rotational Invariant Techniques (ESPRIT) to detect the carrier frequency. This method can be used at lower than normal signal-to-noise ratio's. All results were achieved by the use of computer simulations. http://archive.org/details/detectionofbinar1094523632 Lieutenant Commander, United States Navy Approved for public release; distribution is unlimited.

The number of nonlinear loads connected to power distribution systems is rapidly increasing. These loads disturb the sinusoidal nature of the power distribution systems to which they are connected. Consequently, all connected loads may be affected by even a single nonlinear load. This makes power quality an increasingly important issue, particularly in shipboard applications because equipment malfunction can cause the loss of life. This study primarily develops a modeling technique for use with the harmonic power flow program (HARMFLO) developed by the Electric Power Research Institute. A land based power distribution system is modeled and simulation results are compared to field measurements. Additionally, the causes and effects of harmonic disturbances in power distribution systems are reviewed. In spite of the fact that many parameters of the system being modeled are not precisely known, simulation results are similar to the field measurements. This implies that HARMFLO has applications in isolating existing harmonic disturbances of installed systems and impact assessment for planned systems with one or more nonlinear loads. http://archive.org/details/computernalysiso1094528056 Lieutenant, United States Naval Reserve Approved for public release; distribution is unlimited.

One and two-dimensional system identification and modeling algorithms utilizing multigrid techniques are presented. Finite impulse response (FIR), autoregressive (AR), infinite impulse response (IIR), and 2-D block matrix iterative system modeling algorithms are enhanced and made more efficient using the multigrid methods. The convergence performance of these algorithms is improved with the multigrid techniques. The reduction in the number of iterations required to converge to a solution is realized by forcing the low frequency error components to appear to be at a higher frequency by transferring to a coarser sampling period. Performance comparisons are presented for FIR, AR, IIR, and 2-D block matrix modeling simulations with and without the multigrid techniques employed. http://archive.org/details/iterativesystemm1094526606 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

An automatic control system capable of controlling an unknown nonlinear system is designed using a direct adaptive control scheme, implemented with a Hopfield network. The application of this method to an arbitrary system is discussed in detail and three specific simulation studies are included. These studies include the implementation of the Hopfield network based direct adaptive control system to a linear system, an inverted pendulum, and a nonlinear model of the NPS Autonomous Underwater Vehicle (AUV) with six degrees of freedom. The AUV simulation includes a three dimensional trajectory following algorithm and shows the ability of the Hopfield network to adapt to simultaneous changes in the AUV's depth, speed and course. Additionally, an analog circuit design is proposed which implements the automatic control scheme without the support of a microprocessor. The circuit was set up in SPICE and the simulation results as well as some limitations of the analog circuit implementation of the Hopfield network are presented. http://archive.org/details/ahopfieldnetwork1094526576 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

This thesis investigates the application of backpropagation neural networks as an alternative to adaptive filtering at the NUWES test ranges. To facilitate the investigation, a model of the test range is developed. This model accounts for acoustic transmission losses, the effects of doppler shift, multipath, and finite propagation time delay. After describing the model, the backpropagation neural network algorithm and feature selection for the network are explained. Then, two schemes based on the network's output, signal waveform recovery and binary code recovery, are applied to the model. Simulation results of the signal waveform recovery and direct code recovery schemes are presented for several scenarios. http://archive.org/details/backpropagationn1094526899 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

This thesis investigates the application of adaptive filtering at the NUWES test ranges. Two adaptive algorithms, least-mean-squares and recursive-least squares are studied. To facilitate this investigation, a model of the test ranges was developed. This model accounts for spherical spreading and linear attenuation of the propagated acoustic signals as well as the effects of Doppler shift, multipath, and finite propagation delay time. After describing the model, the adaptive filtering algorithms are briefly explained. Then two schemes of adaptive filtering, adaptive noise cancellation and adaptive line enhancement, are applied to the model. Simulation results of the noise cancellation and tine enhancement schemes are presented for several scenarios. http://archive.org/details/adaptivenoisecan1094527933 Captain, United States Marine Corps Approved for public release; distribution is unlimited.

This thesis investigates the application of artificial neural networks for the compression of image data. An algorithm is developed using the competitive learning paradigm which takes advantage of the parallel processing and classification capability of neural networks to produce an efficient implementation of vector quantization. Multi-Stage, tree searched, and classification vector quantization codebook design techniques are adapted to the neural network design to reduce the computational cost and hardware requirements. The results show that the new algorithm provides a substantial reduction in computational costs and an improvement in performance. http://archive.org/details/datcompressionus1094525801 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

A new two-dimensional data-adaptive algorithm utilizing the iterative Toeplitz approximation method is presented. This algorithm provides a robust and efficient means for accurate estimation of 2-D autoregressive parameters representing spatially variant data arrays. Its convergence performance is comparable to that of the 2-D Recursive Least Squares (RLS) algorithm but is computationally more efficient. The savings in computation is realized by reducing the size of the matrix to be inverted when solving the AR model normal equations. The ability of the algorithm to accurately estimate the model parameters using very small data sets and various windowing schemes are evaluated. Spectral estimates are compared for quarter-plane (QP), nonsymmetric half-plane (NSHP) and combined-quadrant (CQ) regions of support. Additionally, the algorithm is tested in noise cancellation and line enhancement applications using image arrays. This algorithm may be implemented for data-adaptive image processing or coding and for applications requiring 2-D spectral estimation. http://archive.org/details/iterativemethods1094528321 Lieutenant, U.S. Navy Approved for public release; distribution is unlimited.

The prompt and accurate processing of sonar data is essential in undersea warfare. The ability to quickly detect and classify sonar targets is crucial to the performance and survivability of all navy surface ships and submarines. With the advent of neural network technology, new opportunities have arisen which could greatly enhance current sonar target recognition capabilities. The main objective of this research is to demonstrate the practical usage of neural networks in recognizing the acoustic signatures of passive sonar targets using simulated-at-sea conditions. We will review the theory behind neural networks, the problems associated with recognizing acoustic signals in an underwater environment, and we will make a detailed case study of a neural network's performance using test data generated from simulated sonar targets. http://archive.org/details/passivesonartarg1094530962 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

In the past three years a new type of programmable logic device has emerged, The programmable gate array is a new approach to an old problem of trying to implement logic designs in an efficient manner. This thesis explores the implementation of design using the Field Programmable Gate Array (FPGA). In particular, this thesis utilizes the XILINX Logic Cell Array (LCA). This thesis begins by defining the characteristics of the LCA and then define the characteristics of the Small Computer Systems Interface (SCSI) which is used as a design implementation example. The XILINX implementation method is then explored and a complete design implementation study is conducted on the design example. Both Mentor Graphics and Futurenet schematic capture tools are used for design entry. Following design implementation, backannotated design simulation is performed to study the effect of the LCA technology on design performance. The results of this thesis showed that design implemented using this technology performed comparably to other design implementation technologies. Additionally, this implementation method allows design to be completed in a significantly shorter time frame than previously possible. http://archive.org/details/designimplementa1094526451 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

Starting with a least squares formulation of the parameter estimation problem, both fixed data and data-adaptive iterative algorithms are developed. We apply two new techniques, namely diagonal perturbation and multiple partitioning, to existing finite impulse response (FIR) and infinite impulse response (IIR) fixed data matrix splitting algorithms, resulting in improved performance. Also, we extend the fixed data algorithms to the data-adaptive case, and contrast them with FIR and IIR recursive least squares (RLS) algorithms. Computer simulations are used to evaluate the computational effectiveness of the new algorithms. We show the general rate of convergence for the algorithms, evaluate their ability to correctly represent the spectral components of simulated system frequency response in noise, and present system performance, when the order of the model is chosen to be larger than the known system order (over-modeling). http://archive.org/details/iterativemethods1094527629 Captain, United States Army Approved for public release; distribution is unlimited.

Several techniques for estimation of the power spectral density (PSD) , or simply the spectrum, of discretely sampled short data sequences are investigated in this thesis. These techniques are directly based on the fast Fourier transform (FFT) of the data sequence. However, an extrapolated version of the original data sequence will be used instead of the original. The aim is to enhance the resolution of the true spectral components of the signal without adding any false frequency components. Several data extrapolation models are analyzed and their performance at different signal to noise ratios (SNR's) and model orders are examined. http://archive.org/details/spectrumestimati1094527690 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

Travel time of an acoustic signal from transmitter to receiver provides a great deal of information about the ocean environment. Variations in the travel time of the signal may be caused by the changes in the sound speed along the path. Since sound speed is a function of pressure, temperature and salinity, measurement of this parameter in acoustic tomography provides a means to observe ocean fluctuations through the use of inverse techniques. The upcoming Heard Island Experiment will attempt to determine the feasibility of measuring global warming by measuring changes in signal travel time that may be caused by temperature changes in the world's oceans. The signals to be transmitted in this experiment are phase encoded maximal-length sequences of various lengths which are well suited to measurement of travel time. The objectives of this thesis are to provide a software package, in C, that will allow participation as a receiver in this experiment, and to provide a general capability to process any maximal-length sequence, transmitted at any carrier frequency and with any reasonable Doppler. A background on wave propagation, maximal-length sequences, and Doppler processing are presented in this thesis. http://archive.org/details/dopplerprocessin1094534885 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

This thesis treats the topic of multi-dimensional autoregressive (AR) spectral estimation. An iterative algorithm for the solution of toeplitz block- toeplitz matrix equations is presented. This leads to a fast solution of the two dimensional normal equation compared with direct inversion of the autocorrelation matrix. The covariance method is used to estimate the autocorrelation function. Because the resulting matrix is not toeplitz block- toeplitz, a modified iterative algorithm is presented. Quarter-plane and nonsymmetric half-plane support are used, as well as combined quadrant averaging. Results of computer simulation show that in some cases a single iteration is sufficient to produce an acceptable spectral estimate. Because the AR parameters are estimated from previous values, this suggests the possibility to estimate spectral densities of slowly varying random processes. http://archive.org/details/multidimensional1094530666 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

http://archive.org/details/adaptivelgorithm1094522855

This thesis estimates the frequency response of a network where the only data is the output obtained from an Autoregressive-moving average (ARMA) model driven by a random input. Models of random processes and existing methods for solving ARMA models are examined. The estimation is performed iteratively by using the Yule-Walker Equations in three different methods for the AR part and the Cholesky factorization for the MA part. The AR parameters are estimated initially, then MA parameters are estimated assuming that the AR parameters have been compensated for. After the estimation of each parameter set, the original time series is filtered via the inverse of the last estimate of the transfer function of an AR model or MA model, allowing better and better estimation of each model's coefficients. The iteration refers to the procedure of removing the MA or AR part from the random process in an alternating fashion allowing the creation of an almost pure AR or MA process, respectively. As the iteration continues the estimates are improving. When the iteration reaches a point where the coefficients converse the last VIA and AR model coefficients are retained as final estimates. http://archive.org/details/armamodeling1094522912 Lieutenant Junior Grade, Turkish Navy Approved for public release; distribution is unlimited.

We use a block Lanczos algorithm for computing a few of the smallest eigenvalues and the corresponding eigenvectors of a large symmetric matrix rather than computing all the eigenvalue-eigenvector pairs. The basic Lanczos algorithm generates a similar matrix which is block tridiagonal from a given large symmetric matrix. The size of the generated tridiagonal matrix depends upon the number of the smallest eigenvalues to be computed. The result is savings in computations and storage. The block Lanczos algorithm is well -suited for problems involving multiple eigenvalues. In this thesis, we develop the block Lanczos algorithm to estimate the direction-of-arrival (DOA) of a point source based on the observations measured at a linear array of sensors and compare the performance with that of a single vector Lanczos algorithm. The results of the computer simulation experiments conducted with this method are presented and discussed. http://archive.org/details/blocklanczoslgor1094525715 Captain, Korean Army Approved for public release; distribution is unlimited.

http://archive.org/details/twodimensionalut1094526310 NA

http://archive.org/details/digitalsignalpro1094526996 NA NA

A Two-Dimensional Fast Recursive Least Squares (2-D FRLS) algorithm is presented using a geometrical formulation based on the mathematical concepts of vector space, orthogonal projection, and subspace decomposition. By appropriately ordering the 2-D data, the algorithm provides an exact least-squares solution to the deterministic Normal equations. The method is further extended to the general FIR Wiener filter and to ARMA modeling. The size and shape of the support region for both the MA and AR coefficients of the filter can be chosen arbitrarily. The ARMA parameter estimation problem is also considered for the case when the system input is not available. Computer simulations are presented to illustrate the applications of the algorithm for 2-D parameter estimation, system identification and image coding. http://archive.org/details/adaptivetwodimen1094525653 Lieutenant, Portuguese Navy Approved for public release; distribution is unlimited.

The relationship between the structure of a neural network and its ability to perform nonlinear mapping is analyzed. A new algorithm, called the conjugate gradient optimization method, for calculating the weights and thresholds of a neural network is presented. The performance of the conjugate gradient algorithm is then compared to the well known backpropagation method and shown to be more computationally efficient. A neural network using the conjugate gradient algorithm is then applied to three simple examples to demonstrate its signal processing capabilities. The first example illustrates the ability of the neural network to perform classification. The second compares the performance of a one-step linear predictor to a neural network for a nonlinear chaotic time series. The neural network predictor is shown to provide much greater accuracy than its linear counterpart. The final application presented demonstrates the ability of a neural network to perform channel equalization for a nonmininmum phase channel. Its performance is then compared to its linear equivalent. http://archive.org/details/neuralnetworkspp1094526101 Captain, United State Army Approved for public release; distribution is unlimited.

System identification concerns the mathematical modeling of a system based upon its input and output. It allows the development of a mathematical description when all that is available is the result of a process or the output of a system and not the process or system itself. The purpose of this thesis is to develop algorithms for modeling systems as autoregressive-moving-average processes using the method of instrumental variables, a modification of the ordinary least-squares technique, and a multichannel method based upon processing the input and output data by separate infinite-response filters. The methods developed are tested by computer simulation using several second and third-order test cases and the results are presented http://archive.org/details/systemidentifica1094523417 Lieutenant, United States Coast Guard Approved for public release; distribution is unlimited.

Computer simulation studies of a frequency domain adaptive beamforming algorithm are presented. These simulation studies were conducted to determine the multiple broadband target localization capability and the full angular coverage capability of the algorithm. The algorithm was evaluated at several signal-to-noise ratios with varying sampling rates. The number of iterations that the adaptive algorithm took to reach a minimum estimation error was determined. Results of the simulation studies indicate that the algorithm can localize multiple broadband targets and has full angular coverage capability. http://archive.org/details/computersimulati1094522976 Lieutenant, United States Navy Approved for public release; distribution is unlimited.

The classical Levinson-Durbin linear prediction formulas for real valued input sequences are examined and compared to the recently proposed split-Levinson formulas. Both the autoregressive linear predictor model and the adaptive lattice model are used to formulate the new split-Levinson algorithms. A brief introduction to the theory of symmetric polynomials is presented to form the basis of the new algorithms. Computer simulations are used to test and compare the computational accuracy of the new algorithms for AR filter coefficient estimation, parameter estimation for a moving average process, and spectral estimation of sinusoids in white noise. Research results indicate that the new algorithms reduce the number of real multiplications required for a k sub th order AR filter problem by one-half, and they are applicable to both the extended Prony method of spectral estimation of moving average parameters. Keywords: Text processing, Word processing, Theses. (kr) http://archive.org/details/asplitlevisonppr1094523403 Lieutenant Commander, United States Navy Approved for public release; distribution is unlimited.

Subspace methods of solving spectral estimation and direction of arrival (DOA) problems involve finding the eigenvalues and eigenvectors of correlation matrices. Using the Lanczos algorithm some of the extreme eigenvalues and eigenvectors can be approximated without requiring the entire matrix decomposition theoretically saving many computations. This thesis develops a model and a form of the Lanczos algorithm to solve the DOA problem. The relationship of the number of eigenvectors used to the accuracy of the results in a low signal to noise ratio example are examined. http://archive.org/details/doaestimationbye1094527000 Lieutenant Commander, United States Navy Approved for public release; distribution is unlimited.

http://archive.org/details/adaptivearmalatt1094522856

In computer vision many techniques have been developed for object recognition. The affine invariant matching algorithm proposed by Hummel and Wolfson (1988) is a new and interesting method. Under affine invariant transformation, objects with translation, rotation, scale changes, and, or even partial occlusion will have the same or similar coefficients. However, some serious problems exist in the original algorithm. This thesis begins with the discussion of the affine transformation. The shortcomings that can occur in this method such as the basis instability, the collision of hash table, and the noise sensitivity will be discussed. Among them the noise sensitivity is a serious problem. This can always cause the recognition procedure to fail. In this thesis an improved affine invariant matching algorithm was developed to overcome the noise problem and other disadvantages of the original algorithm. The area test criteria were adopted to avoid the numerical instability problem. The modified hashing structure using a special hash function was implemented to achieve faster accessing and voting. In the recognition procedure, the partial voting technique with the consideration of false peaks from the voting array highly enhanced the noise tolerance of the algorithm. Finally, the results obtained from the improved algorithm clearly showed better performance than those of the original algorithm. http://archive.org/details/affineinvariantm1094526852 Lieutenant Commander, Taiwan Republic of China Navy Approved for public release; distribution is unlimited.