Personal communications has produced a demand for low-Earth orbit satellite (LEOSAT) implementation. A proposed LEOSAT link at VHF would provide paging and radio-location services to subscribers in automobiles. Link performance has traditionally been determined by thermal background noise produced internally as well as externally to the radio receiver. Most recently, however, man-made noise has become the dominant source of link interference at VHF frequencies and below. This thesis examines the statistical behavior of man-made noise through a series of in situ measurements, and implements a statistical-physical mathematical model to provide bit-error rate (BER) predictions for system designers. The model is also applied in an attempt to scale the statistics for bandwidths other than the noise measurement receiver. Man-made noise is inherently impulsive, due to switching processes in the AC power distribution system. Gaussian noise models are therefore inappropriate for the description of man-made noise statistics. A model by Middleton was developed around the Poisson mechanism to accurately predict the amplitude probability distribution (APD) of the received noise. Unlike empirical models, the constants employed by the Middleton model posses physical significance. / Master of Science

The present paper introduces one very efficient and flexible time-domain analysis technique to predict the kinds of reflections and crosstalk. Numerical results show that this technique is indeed efficient and accurate in the transient analysis of a general multiple line system. Furthermore, this algorithm can be eventually coded in a form of a subroutine compatible with any standard CAD program, such as SPICE.

A dissertation submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of requirements for the degree of Master of Science. Johannesburg, July 14, 2015. / Hyperspectral images nd application in many areas of modern society, we use them for land surveying, core sample analysis, in the conservation and forestry industries and many more. A major problem in hyperspectral images is how to deal with noise. Many methods that analyse hyperspectral images either need clean images or accurate estimations of the noise statistics in the images. The goal of this dissertation is to present and compare methods for statistic estimation and noise removal. We use an arti cial hyperspectral image to study some existing methods and develop some new ones based on existing methods, speci cally the BM3D algorithm. We test methods that estimate the level of the noise present in an image, methods that estimate the structure of the noise and methods that remove noise. We analyse all the methods under an additive noise model and consider spectrally correlated and uncorrelated noise. Within our investigations we investigate di erent types of correlation. We will show the strengths that the various methods have and establish a way to approach treating a hyperspectral image with no information beyond the image itself. Using our observations and insights from the experiments on the arti cial data we analyse some radiance data from the AVIRIS instrument. We show that the additive signal independent part of the noise is small but not negligible. We also show some evidence for the structure of the noise in the AVIRIS instrument.

Electrochemical noise refers to the spontaneous fluctuations in potential and current that can be observed on a corroding metal. The use of electrochemical noise for obtaining information on the corrosion process generates much interest in research fields. One important application is the measurement of corrosion rate. This can be achieved using the electrochemical noise of a pair of electrically coupled corroding metals to obtain an estimate of electrochemical impedance - an abstract quantity that reflects various aspects of the corrosion process.There are a number of problems associated with estimation of impedance information from the electrochemical noise data, particularly regarding data pre-treatment, accuracy and precision. In addition, the present methods are incomplete: current literature does not offer information regarding the phase of the impedance; and assumptions regarding symmetry of an electrode pair cannot be tested without additional measurements.The thesis addresses the above mentioned problems. Specifically,analysis of the impedance estimation process is given to determine how precision can be affected by various factors;a novel signal processing technique is described that is shown to yield a local optimum precision;the application of the proposed signal processing to time varying systems is demonstrated by use of a time varying, frequency dependent impedance estimate;a technique for recovering phase information, given certain conditions, is suggested so that Nyquist impedance diagrams can be constructed; anda technique for testing the symmetry of a coupled pair of corroding metals is described.An integral part of electrochemical noise analysis is the software used for numerical computation. The Matlab package from MathWorks inc. provides an extensible platform for electrochemical noise analysis. Matlab code is provided in Appendix A to implement ++ / much of the theory discussed in the thesis.Impedance analysis and many other electrochemical corrosion monitoring techniques are primarily used for uniform corrosion, where the corrosion patterns occur uniformly over the exposed surface. In order to map localised corrosion, where the corrosion is typically concentrated within a small area, a wire beam electrode can be used. A wire beam electrode is a surface that is divided into a matrix of mini-electrodes so that the corrosion rate at different points can be monitored. However, manual connection of each mini-electrode to the measurement device can prove cumbersome. The final chapter of this thesis describes the design and testing of specialised multiplexing hardware to automate the process.In general, the thesis shows that by careful conditioning of the electrochemical noise prior to analysis, many of the problems with the technique of impedance estimation from the electrochemical noise data can be overcome. It is shown that the electrochemical noise impedance estimation can be extended to encompass a time varying, frequency dependent quantity for studying dynamic systems; that phase information can be recovered from electrochemical noise for the purpose of constructing Nyquist impedance diagrams; and that asymmetric electrodes can be detected without requiring additional measurements.

Alfred Zacharias, Louis D. Smullin. / Based on a thesis, M.I.T. Dept. of Electrical Engineering, May 25, 1959. / Bibliography: p. 23. / Army Signal Corps Contract DA36-039-sc-78108. Dept. of the Army Task 3-99-20-001 and Project 3-99-00-000.

Shigebumi Saito. / "August 22, 1957. " / Bibliography: p. 50. / Army Signal Corps Contract DA36-039-sc-64637 Dept. of the Army Task 3-99-06-108 Project 3-99-00-100

Hospital sound environments are complex and hard to understand. One of the most important factors in these environments is the effective communication between staff members in regards to patient care and successful communication depends in part on the hospital’s sound environment. In this study, objective sound measurements as well as occupant perceptive data were collected at three hospitals. Sound pressure levels; including maximum, peak, minimum and equivalent levels were recorded in these hospitals, in addition to active impulse response measurements. Acoustic descriptors of the sound environment such as spectral content, level distributions, energy decay and temporal patterns were examined. The perception of the hospital soundscape (sound environment) was evaluated through surveys of the staff, patients and visitors to units. It was found that noise levels in all patient rooms and work areas were significantly higher than guidelines laid out in previous literature and by professional organizations. This work contributes to the field by broadening the metrics used to quantify hospital acoustic environments. In addition, this work added to the field by providing the most rigorous acoustic field measurement set published to date. This was done to create an accurate portrayal of the hospital soundscape environment.

No description available.

Speech intelligibility tests using three headset systems, two passive and one with active noise reduction technology, were conducted. Nine listeners, six males and three females, responded to 180 50-word Modified Rhyme Word Lists presented under two noise conditions, two conditions of bilateral phase reversal of the speech signal, and the three headset systems. These three independent variables were studied in a full factorial within-subjects design. The noise conditions were pink noise and recordings from the troop carrying compartment of a U.S. Army M-2 Bradley Fighting Vehicle (tank noise having substantial low-frequency energy). Phase reversal of the speech signal was also tested for its effect on speech intelligibility. The three headset systems were the David Clark Noise Attenuating Aviation Headset, and the Bose Aviation Headset used in both its passive and active mode. Active noise reduction technology employs the physical principle of constructive and destructive interference by creating signals which are identical to the noise under the earmuff but out of phase by 180 degrees. This technology is most effective at frequencies below 1000 Hz. Results showed that the Bose headset in its active mode required a significantly higher speech-to-noise ratio in both noise environments than the two passive headset systems to achieve the 70% level of speech intelligibility, the dependent variable in this study. The mean speech level under the Bose headset in its active mode was 2.8 dB higher in tank noise and 3.5 dB higher in pink noise than under the David Clark headset. The Bose headset in the active mode, however, provided a greater degree of broadband attenuation especially in the lower frequency range, e.g. < 630 Hz. Phase reversal proved to be of no benefit to performance in either noise environment. The pink noise proved to be the harsher environment for speech intelligibility than did the tank noise, primarily due to the increased levels in the speech bandwidth. Articulation Index scores for the three headset systems evidenced that the differences in performance among the three headset systems were in part the result of better earphone response characteristics in the principal speech bandwidth (600 - 4000 Hz) in conjunction with the strong attenuation performance of the David Clark headset in that same range. It is suggested that, in the absence of better earphone response characteristics and a broader bandwidth of active noise reduction performance, speech intelligibility is no better under active noise reduction headsets than under quality passive headsets. Further, the levels of speech intelligibility attained by active noise reduction headsets are at a cost of increased signal strength and higher purchase price. / Ph. D.

The acoustic emission from the combustion of turbulent mixtures of several gaseous fuels and oxygen was measured to determine the effect of burner diameters, flow velocities and reactant chemistries. The data are presented in the form of acoustic pressure spectra and over-all sound pressure levels. The experiment consisted of a geometrically simple pre-mixed burner with five interchangeable nozzles whose diameters varied from 0.117 cm to 0.541 cm. The fuels examined in the experiment were acetylene, methane and propane. Three flow velocities were tested for each fuel. Methane was tested within a velocity range of 10 m/s to 25 m/s while propane was examined within a range of 20 m/s to 40 m/s. Acetylene was tested at somewhat higher velocities ranging from 50 m/s to 100 m/s. The resulting data base consisted of 23 unique test conditions. The data were evaluated based on peak frequency and overall sound pressure level. The experimental data are also compared to a predictive model. / Master of Science