The matrix-vector multiplier is an important building block in optical information processing architectures, examples of which are correlators for pattern recognition, associative memories, and neural networks. Such architectures are most suitable for implementation by optics due to the ease in realizing dense interconnections optically. The success of the implementation partially relies on the quality of the SLM used to record the information for processing. Limited dynamic range for the representation of the data recorded is a common drawback suffered by most commercially available devices. In this thesis, the importance of the dynamic range of the device on the performance of the implementation is investigated. The effect of limited dynamic range on the signal to noise ratio, probability of error, capacity, and training of various forms of matrix-vector multipliers are addressed. Through the use of theoretical analyses, computer simulations, and optical experiments, it will be shown that a large dynamic range is not essential in most applications. Specifically, it is shown that only one bit of dynamic range, i.e. two gray levels, for the representation of each data point, results in acceptable loss in performance.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Psaltis, Demetri}, } @phdthesis{10.7907/65XX-PQ66, author = {Williams, Wyman Lee}, title = {Computer-Aided Measurement of Microwave Circuits}, school = {California Institute of Technology}, year = {1989}, doi = {10.7907/65XX-PQ66}, url = {https://resolver.caltech.edu/CaltechETD:etd-02162007-080706}, abstract = {Instruments that measure the scattering parameters of microwave circuits generally have large systematic errors due to unavoidable parasitics in the instruments. These errors can be modeled analytically, however, and removed through a calibration procedure. A personal computer is well suited to the performance of the required calculations. Combining a personal computer with a microwave network analyzer results in a flexible and accurate automatic instrument. Two such automatic network analyzers are presented here. A new type of network analyzer, known as a sampled-line network analyzer is presented. It is an extension of the six-port network analyzer concept developed at the National Bureau of Standards. It is a particularly simple implementation and shows promise for the construction of relatively low-cost microwave network analyzers. The sampled-line network analyzer is analyzed theoretically and several experimental versions of it are presented. Another personal computer-controlled network analyzer is presented in which a Pascal program automates an HP 8410 network analyzer. The result is an instrument which can measure S-parameters from 0.5 to 18 GHz with a measurement error vector ranging in magnitude from 0.01 in the low frequency range to about 0.03 at 18 GHz.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Rutledge, David B.}, } @phdthesis{10.7907/MM2J-E556, author = {Sánchez Peña, Ricardo Salvador}, title = {Robust Analysis of Feedback Systems with Parametric and Dynamic Structured Uncertainty}, school = {California Institute of Technology}, year = {1989}, doi = {10.7907/MM2J-E556}, url = {https://resolver.caltech.edu/CaltechETD:etd-02012005-084251}, abstract = {This thesis presents the first general program implementation of the algorithm by deGaston and its generalization by Sideris and deGaston to compute the Multivariable stability margin or Structured singular value of a feedback system under real (independent or related) parametric uncertainty. An improved implementation of the algorithm mentioned above is also considered, which simplifies significantly the code and increases the computational speed. The latter also allows a simple and fast analysis by just checking the extreme values of the set of parameters, with a high probability of achieving the actual stability margin; this being supported by an intense statistical analysis performed at the end of this thesis.
A great deal of work has recently been done related to this class of uncertain systems initiated by the well known theorem of Kharitonov. A connection is made in Chapter 4 between these procedures and the above ones in terms of generality of the class of uncertain polynomials considered. A theorem characterizing the set of polynomials whose robust stability can be determined by a finite number of tests is addressed. Sufficient conditions to determine when the latter conditions apply are also given, which in some cases can considerably simplify the analysis. In particular cases, polynomials with related uncertain parameters can be treated in the same way as independent parameters as shown in two examples.
The main part of this thesis is concerned with the analysis of more general type of uncertainties. In particular, the analysis of robust stability for the case when unstructured dynamic uncertainty is combined with real parametric uncertainty is treated in Chapter 5. This can also be applied in the analysis of robust performance for plants with parametric uncertainty. Chapter 6 generalizes the latter to the most general case in which structured dynamic and real parametric uncertainty appear simultaneously in the plant. A computational scheme is given in both cases which uses the algorithm mentioned in the first part and is applied to several examples.
At the end, an example of the robust analysis of an experimental aircraft demonstrates how a practical situation can be handled by this procedure.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Sideris, Athanasios}, } @phdthesis{10.7907/7ytp-0932, author = {So, Frank Cheung Tao}, title = {Diffusion Barriers for VLSI Applications}, school = {California Institute of Technology}, year = {1988}, doi = {10.7907/7ytp-0932}, url = {https://resolver.caltech.edu/CaltechETD:etd-02012007-110846}, abstract = {This thesis is concerned with diffusion barriers in contact structures to semiconductors. Diffusion barriers are indispensable in present contact technologies to preserve device characteristics from the influence of metal-semiconductor interaction during post-metallization processing.
The absence of grain boundaries makes amorphous W-Zr and Ni-W alloy barriers very attractive for high temperature applications. Nevetheless, in the presence of an adjoining metal layer such as Al, these amorphous barriers are chemically dissociated to form compounds with the metal. The usefulness of these barrier in VLSI metallization schemes is severely limited by their high reactivity with Al within normal processing temperature cycles. Such thermal instability can be removed, however, by adding nitrogen to the barrier layers during sputter deposition.
Becoming aware of the beneficial effects of nitrogen incorporation, we investigate the performance of nitrogen-doped W barrier films in various contact configurations. Reactively sputtered amorphous and polycrystalline W-N layers are demonstrated to be excellent diffusion barriers against interdiffusion between Si-Al, Si-Ag, GaAs-Ag, and GaAs-Au. A novel idea of utilizing W-N as an interconnect in CMOS fabrication is also discussed.
Conducting transition metal oxides emerge as a new class of diffusion barriers. Metallic Mo1-xOx films are deposited by reactive sputtering a Mo target in controlled O₂/Ar ambients. These Mo1-xOx barriers can effectively protect Si n⁺-p shallow junctions from Al spiking even beyond the eutectic temperature of Si-Al. RuO₂ films are also found to be equally good in suppressing Si-Al interdiffusion. The present study clearly shows that Mo1-xOx and RuO₂ barriers are the most outstanding performers among all the passive barriers that have been explored so far.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Nicolet, Marc-Aurele}, } @phdthesis{10.7907/B9FG-6H18, author = {Rebeiz, Gabriel M.}, title = {Monolithic Millimeter-Wave Two-Dimensional Horn Imaging Arrays}, school = {California Institute of Technology}, year = {1988}, doi = {10.7907/B9FG-6H18}, url = {https://resolver.caltech.edu/CaltechETD:etd-11082007-111802}, abstract = {A new monolithic millimeter-wave two-dimensional horn imaging array is presented. In this configuration, a dipole is suspended in an etched pyramidal cavity on a 1-µm silicon-oxynitride membrane. This approach allows ample space for low-frequency interconnections, while still maintaining efficient diffraction-limited imaging. The fabrication procedure of the horn array and the deposition parameters of the membrane layer are presented in detail. The array is analysed rigorously, by approximating the horn antenna by a structure of multiple rectangular waveguide sections. Pattern measurements at 93 GHz and 242 GHz agree well with the theory. The results show that horn antennas with an opening between 1.0λ and 1.5λ have high aperture efficiencies and would match well appropriate imaging systems. Also, a new wideband log-periodic antennas is integrated on a thin membrane and tuned by a back plane reflector. The antenna patterns are measured at 167 GHz, 370 GHz and 700 GHz, and the effect of the back-plane reflector is investigated at 370 GHz.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Martel, Hardy Cross}, } @phdthesis{10.7907/cn89-3127, author = {Smyth, Padhraic}, title = {The Application of Information Theory to Problems in Decision Tree Design and Rule-Based Expert Systems}, school = {California Institute of Technology}, year = {1988}, doi = {10.7907/cn89-3127}, url = {https://resolver.caltech.edu/CaltechETD:etd-02022007-103549}, abstract = {This thesis examines the problems of designing decision trees and expert systems from an information-theoretic viewpoint. A well-known greedy algorithm using mutual information for tree design is analysed. A basic model for tree design is developed leading to a series of bounds relating tree performance parameters. Analogies with prefix-coding and rate-distortion theory lead to interesting interpretations and results. The problem of finding termination rules for such greedy algorithms is discussed in the context of the theoretical models derived earlier, and several experimentally observed phenomena are explained in this manner. In two classification experiments, involving alphanumeric LEDS and local edge detection, the hierarchical approach is seen to offer significant advantages over alternative techniques.
The second part of the thesis begins by analysing the difficulties in designing rule-based expert systems. The inability to model uncertainty in an effective manner is identified as a key limitation of existing approaches. Accordingly, an information-theoretic model for rules and rule-based systems is developed. From a simple definition of rule information content, the ability to specialise and generalise (akin to cognitive processes) in a quantitative manner is demonstrated. The problem of generalised rule induction is posed and the ITRULE algorithm is described which derives optimal rule sets from data. The problem of probabilistic updating in inference nets is discussed and a new maximum-likelihhod rule is proposed based on bounded probabilities. Utility functions and statistical decision theory concepts are used to develop a model of implicit control for rule-based inference. The theory is demonstrated by deriving rules from expert-supplied data and performing backward and forward chaining based on decision-theoretic criteria. The thesis concludes by outlining the many problems which remain to be solved in this area, and by briefly discussing the analogies between rule-based inference nets and neural networks.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Goodman, Rodney M.}, } @phdthesis{10.7907/2E2R-Z765, author = {Dukelow, Robert Alexander}, title = {An experimental investigation of very low frequency semiconductor noise}, school = {California Institute of Technology}, year = {1974}, doi = {10.7907/2E2R-Z765}, url = {https://resolver.caltech.edu/CaltechETD:etd-12052005-155759}, abstract = {NOTE: Text or symbols not renderable in plain ASCII are indicated by […]. Abstract is included in .pdf document.
This thesis represents an attempt to more accurately characterize the statistics of the very low frequency noise of a particular semi conductor device (a grounded input bipolar integrated operational amplifier) than has previously been achieved. Power spectral density estimates are obtained for frequencies ranging from 250 Hz to […] Hz. These estimates are based on data recorded over an uninterrupted period of approximately 1 year (355.9 days). Relatively high sample rates are maintained for each source, allowing a high degree of accuracy in these estimates without resorting to the questionable process of averaging the estimates for a number of noise sources [1]. The high sample rate also allows for a reduction in the errors due to aliasing by the use of digital filtering techniques. Spectral density estimates for six separate noise sources are presented.
The preliminary objectives of the experiment were to search for a break in the […] spectral density component of semiconductor noise and to attempt to establish the “true” value of [alpha] if a unique value exists. Very long time constant “popcorn” noise (ignored by one investigator because it is “as natural as flicker noise itself” [1, page 69]), proved to be an obstacle in measuring the “pure” l/f noise process and may not only be the reason for the wide range of values for a reported in the literature but, as suggested by some researchers [2], may be the cause of flicker noise. “Popcorn” noise was observed with apparent time constants greater than […] seconds. The statistics of the observed “popcorn” noise were investigated and showed good agreement with the results of J. N. Puckett, Jr. [3] (who worked primarily with popcorn noise having time constants on the order of a few milliseconds) except for a few cases in which the waveform resembled that which might be expected if one popcorn component were to modulate another. The work of Puckett was also extended in that a test for burst waiting time dependency was performed. No evidence of dependency could be found. Popcorn noise components which were large enough to be clearly identified were removed in the time domain. This technique was found to be quite useful in improving the spectral estimates.
As in previous experiments, power supply regulation and temperature control were found to be essential. In this experiment, however, the temperatures of the noise sources and the power supply fluctuations were measured concurrently with the noise data so that their contribution to the total observed noise could be more accurately ascertained.
Neither a break frequency in the […] trend nor a unique [alpha] were established, although the values of [alpha] obtained were all slightly larger than but closer to 1.0 than most reported values. A nearly periodic component (with a period of 1 year) was also observed in two of the noise sources which were contained in a plastic integrated circuit package. Since this component cannot be explained by temperature or power supply voltage it is conjectured that other external parameters such as humidity or barometric pressure (or even cosmic radiation) may account for some of the observed noise, although no such component could be observed in the other four noise sources (which were housed in metal packages).}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Martel, Hardy Cross}, } @phdthesis{10.7907/MDV9-W982, author = {Caloyannides, Michael Akylas}, title = {A mathematical and experimental investigation of microcycle spectral estimates of semiconductor flicker noise}, school = {California Institute of Technology}, year = {1972}, doi = {10.7907/MDV9-W982}, url = {https://resolver.caltech.edu/CaltechTHESIS:04042016-152719186}, abstract = {The experimental portion of this thesis tries to estimate the density of the power spectrum of very low frequency semiconductor noise, from 10-6.3 cps to 1. cps with a greater accuracy than that achieved in previous similar attempts: it is concluded that the spectrum is 1/fα with α approximately 1.3 over most of the frequency range, but appearing to have a value of about 1 in the lowest decade. The noise sources are, among others, the first stage circuits of a grounded input silicon epitaxial operational amplifier. This thesis also investigates a peculiar form of stationarity which seems to distinguish flicker noise from other semiconductor noise.
In order to decrease by an order of magnitude the pernicious effects of temperature drifts, semiconductor “aging”, and possible mechanical failures associated with prolonged periods of data taking, 10 independent noise sources were time-multiplexed and their spectral estimates were subsequently averaged. If the sources have similar spectra, it is demonstrated that this reduces the necessary data-taking time by a factor of 10 for a given accuracy.
In view of the measured high temperature sensitivity of the noise sources, it was necessary to combine the passive attenuation of a special-material container with active control. The noise sources were placed in a copper-epoxy container of high heat capacity and medium heat conductivity, and that container was immersed in a temperature controlled circulating ethylene-glycol bath.
Other spectra of interest, estimated from data taken concurrently with the semiconductor noise data were the spectra of the bath’s controlled temperature, the semiconductor surface temperature, and the power supply voltage amplitude fluctuations. A brief description of the equipment constructed to obtain the aforementioned data is included.
The analytical portion of this work is concerned with the following questions: what is the best final spectral density estimate given 10 statistically independent ones of varying quality and magnitude? How can the Blackman and Tukey algorithm which is used for spectral estimation in this work be improved upon? How can non-equidistant sampling reduce data processing cost? Should one try to remove common trands shared by supposedly statistically independent noise sources and, if so, what are the mathematical difficulties involved? What is a physically plausible mathematical model that can account for flicker noise and what are the mathematical implications on its statistical properties? Finally, the variance of the spectral estimate obtained through the Blackman/Tukey algorithm is analyzed in greater detail; the variance is shown to diverge for α ≥ 1 in an assumed power spectrum of k/|f|α, unless the assumed spectrum is “truncated”.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Martel, Hardy Cross}, } @phdthesis{10.7907/ymat-1n64, author = {Adegbola, Mashood Olayide}, title = {Alias-Free Spectral Estimation of Stochastic Processes}, school = {California Institute of Technology}, year = {1971}, doi = {10.7907/ymat-1n64}, url = {https://resolver.caltech.edu/CaltechTHESIS:10112017-132121017}, abstract = {
A scheme for the practical estimation of power spectrum from randomly-timed samples is proposed and investigated for wide-sense stationary point processes. The sampling process {tn} is assumed to be stationary point process statistically independent of the sampled process X(t). Stationarity of {tn} admits that joint statistics of tk, tk+n do not depend on k. Closed form analytical formulae are derived for the spectral window Qm(f) and for cov{S^(fr), S^(fq)}, var{S^(fr)} for the particular case of independent identically distributed sampling intervals. Results confirm the alias-free character of the Poisson sampling scheme even for non-bandlimited spectra. It is shown further that for Gaussian processes with very smooth spectra Poisson sampling process can yield more reliable estimates (i.e., with a smaller variance) than the well known method of periodic sampling.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Martel, Hardy Cross}, } @phdthesis{10.7907/D816-7C55, author = {Puckett, Jason Niles, Jr.}, title = {An Electrical and Statistical Study of Burst Noise}, school = {California Institute of Technology}, year = {1971}, doi = {10.7907/D816-7C55}, url = {https://resolver.caltech.edu/CaltechTHESIS:04042018-105756176}, abstract = {Burst noise is a normally undesirable phenomenon occasionally found in bipolar semiconductors and other current carrying devices. It is an electrical fluctuation which exhibits itself as one or more rectangular waveforms possessing constant amplitude but random pulse duration. The experimental portion of this study relates only to burst noise in bipolar transistors and operational amplifiers.
Burst noise is not Gaussian as are the more common fluctuations in semiconductors. That fact was established by estimation of the amplitude distribution, a technique found to be sensitive in the detection of burst noise obscured by quantities of conventional noise.
The amplitude of burst noise varies with the parameters of base-emitter, voltage temperature and source resistance. An exponential increase of amplitude with Vbe and a lack of dependence on collector voltage implied that the noise originates in the base-emitter junction. A noise magnitude linearly proportional to source resistance over several decades leads one to infer the equivalent circuit of a current source between base and emitter. Current amplitudes of 10-10 to nearly 10-6 ampere p-p were observed.
Burst noise pulse durations were found as brief as 10 µsec and as long as some 29 hours; neither an upper nor a lower bound was established. The two noise states (high and low, in the rectangular waveform) were treated separately in the duration experiments. Careful measurements on the relative frequency with which the pulse occurred gave duration probability densities of 1/τ e -t/τ for each state. That density also applies to a single particle alternately being trapped and escaping and is consistent with the physical theory due to Mead and Whittier relating burst noise to trapping phenomena. Measurements on noise pulse durations in both states as a function of Vbe lent support to the theory and indicated both trapping and recombination-generation centers were present in samples examined. Another theory, due to Leonard and Jaskolsky, was found inconsistent with the evidence for burst noise’s origin in the base-emitter junction. Duration versus temperature dependence indicated activation energies of roughly .5 eV.
Although suggestions in the literature for the power spectrum of burst noise have been inconsistent, digital spectral estimation and judicious use of a wave analyzer showed the spectrum to be flat at low frequencies and to fall as 1/f2 at higher ones. Proceeding only from the measured pulse duration probability density, the power spectrum was deduced on theoretical grounds for the first time. The method entailed the derivation of burst noise’s autocorrelation function which, when Fourier transformed, yielded
S(w) = 2/(τ1 + τ0)[(1/τ1 + 1/τ0)2 + w2]
where τ1 and τ0 are the average durations in the two noise estates. The expression proved consistent with experiment.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Martel, Hardy Cross}, } @phdthesis{10.7907/F0TV-2421, author = {Davey, Theodore Herbert}, title = {The conjugate representation of signal envelope}, school = {California Institute of Technology}, year = {1970}, doi = {10.7907/F0TV-2421}, url = {https://resolver.caltech.edu/CaltechTHESIS:08202015-153803402}, abstract = {
The problem of the representation of signal envelope is treated, motivated by the classical Hilbert representation in which the envelope is represented in terms of the received signal and its Hilbert transform. It is shown that the Hilbert representation is the proper one if the received signal is strictly bandlimited but that some other filter is more appropriate in the bandunlimited case. A specific alternative filter, the conjugate filter, is proposed and the overall envelope estimation error is evaluated to show that for a specific received signal power spectral density the proposed filter yields a lower envelope error than the Hilbert filter.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Martel, Hardy Cross}, } @phdthesis{10.7907/wfs6-ej43, author = {Green, Richard Rutherford}, title = {Decoding cosets of first-order Reed-Muller code}, school = {California Institute of Technology}, year = {1969}, doi = {10.7907/wfs6-ej43}, url = {https://resolver.caltech.edu/CaltechTHESIS:03182016-151447208}, abstract = {Proper encoding of transmitted information can improve the performance of a communication system. To recover the information at the receiver it is necessary to decode the received signal. For many codes the complexity and slowness of the decoder is so severe that the code is not feasible for practical use. This thesis considers the decoding problem for one such class of codes, the comma-free codes related to the first-order Reed-Muller codes.
A factorization of the code matrix is found which leads to a simple, fast, minimum memory, decoder. The decoder is modular and only n modules are needed to decode a code of length 2n. The relevant factorization is extended to any code defined by a sequence of Kronecker products.
The problem of monitoring the correct synchronization position is also considered. A general answer seems to depend upon more detailed knowledge of the structure of comma-free codes. However, a technique is presented which gives useful results in many specific cases.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Martel, Hardy Cross}, } @phdthesis{10.7907/XE6F-BM96, author = {Titsworth, Robert C.}, title = {Correlation properties of cyclic sequences}, school = {California Institute of Technology}, year = {1962}, doi = {10.7907/XE6F-BM96}, url = {https://resolver.caltech.edu/CaltechETD:etd-09092005-130635}, abstract = {
In the unconstrained channel with additive Gaussian noise, where the optimum detector is based on correlation or matched filters, the quality of a code can be expressed as a function of the correlation values between code words. For a cyclic-sequence code, optimality reduces to a criterion to be met by the autocorrelation function of the sequence. Methods are presented here for determining cyclic sequences with given correlation properties.
When the amount of equipment in the receiver is limited, matched filtering is no longer the optimal detection scheme. A better system, as is shown here, is one which, by the use of a Boolean function, combines several “component” sequences to generate the transmitted signal; the receiver consists of filters matched to each component. The logic, the number of components, the requirements of the component sequences to optimize the system, and a general method for treating Boolean logics are given in this work.
}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Golomb, Solomon W. and Martel, Hardy Cross}, } @phdthesis{10.7907/EB99-8T09, author = {Goldstein, Richard M}, title = {Radar exploration of Venus}, school = {California Institute of Technology}, year = {1962}, doi = {10.7907/EB99-8T09}, url = {https://resolver.caltech.edu/CaltechTHESIS:07212010-095208070}, abstract = {The feasibility of studying Venus by radar was demonstrated on the tenth of March, 1961, when an echo from Venus was first detected in real time. The spectrum of the echo may be expected to be broadened by the doppler shifts produced by any rotation Venus might have. In order to measure this broadening, a novel technique for accurate spectral estimation of this extremely weak signal was devised, analyzed and implemented. The results of the measurements indicate that Venus rotates only once for each of its years, so that one hemisphere always faces the sun. In addition, the spectral measurements together with some polarization measurements enable one to infer some scattering properties of the surface. Time of flight measurements were also made, using a cross-correlation technique. These, compared with standard astronomical tables, provide a new value for the Astronomical Unit. The standard deviation of these measurements is only a few parts in 10^7.}, address = {1200 East California Boulevard, Pasadena, California 91125}, advisor = {Martel, Hardy Cross}, }