1,721,012 research outputs found
Cognon Neural Model Software Verification and Hardware Implementation Design
Full text linkLittle is known yet about how the brain can recognize arbitrary sensory patterns within milliseconds using neural spikes to communicate information between neurons. In a typical brain there are several layers of neurons, with each neuron axon connecting to ∼104 synapses of neurons in an adjacent layer. The information necessary for cognition is contained in theses synapses, which strengthen during the learning phase in response to newly presented spike patterns. Continuing on the model proposed in "Models for Neural Spike Computation and Cognition" by David H. Staelin and Carl H. Staelin, this study seeks to understand cognition from an information theoretic perspective and develop potential models for artificial implementation of cognition based on neuronal models. To do so we focus on the mathematical properties and limitations of spike-based cognition consistent with existing neurological observations. We validate the cognon model through software simulation and develop concepts for an optical hardware implementation of a network of artificial neural cognons
Cognon Neural Model Software Verification and Hardware Implementation Design
Full text linkLittle is known yet about how the brain can recognize arbitrary sensory patterns within milliseconds using neural spikes to communicate information between neurons. In a typical brain there are several layers of neurons, with each neuron axon connecting to ∼104 synapses of neurons in an adjacent layer. The information necessary for cognition is contained in theses synapses, which strengthen during the learning phase in response to newly presented spike patterns. Continuing on the model proposed in "Models for Neural Spike Computation and Cognition" by David H. Staelin and Carl H. Staelin, this study seeks to understand cognition from an information theoretic perspective and develop potential models for artificial implementation of cognition based on neuronal models. To do so we focus on the mathematical properties and limitations of spike-based cognition consistent with existing neurological observations. We validate the cognon model through software simulation and develop concepts for an optical hardware implementation of a network of artificial neural cognons
Manifold-Based System for Passive-Active Spectrum Sharing
Full text linkA feasibility study for an automated scheme for spectrum sharing between passive and active users is presented. The needs of spectrum users can be represented by manifolds in a Euclidean hyperspace called electrospace which has 7 dimensions: frequency (f); Cartesian coordinates (x,y,z); angular coordinates (θ,ø), and time (t). The entire globe is tessellated into geographical areas containing spectrum users, called user domains. Each user domain is recursively tessellated into smaller user domains, or subdomains. A computer cluster, or broker, in each smallest subdomain performs the calculations necessary to determine if a particular user in the subdomain experiences interference. Throughout this thesis, the Chicago Loop (area of 4.09 km2, population ~ 21, 000) is taken to be the representative example of a smallest subdomain.
Within each subdomain, the number of users served by a broker is reduced to a manageable number by the process of culling. There are three orders of culling. In first-order culling, subdomain pairs without line of sight and not close enough to mutually interfere are culled, or removed from further consideration for interference calculations. In second-order culling, within each subdomain an intersection test of the electrospace manifolds of all user pairs is performed. User pairs whose manifolds do not intersect are culled. In third-order culling a Friis calculation is performed for all remaining user pairs. The output of third-order culling is an RFI flag bit for each user indicating whether interference is present or not. The computational complexity of first-, second-, and third-order culling calculations was determined.
Three representative user classes will be discussed: WiFi access points, Terminal Doppler Weather Radars, and passive EESS satellites. The manifold descriptor language (MDL) for each of the three user classes was described. The computational complexity of broker calculations to determine electrospace parameters from the MDL was determined. Using this complexity and the complexity of culling calculations, the total computational requirements for a broker in a representative subdomain is determined in GFLOPS (Giga Floating Point Operations Per Second).</p
Manifold-Based System for Passive-Active Spectrum Sharing
Full text linkA feasibility study for an automated scheme for spectrum sharing between passive and active users is presented. The needs of spectrum users can be represented by manifolds in a Euclidean hyperspace called electrospace which has 7 dimensions: frequency (f); Cartesian coordinates (x,y,z); angular coordinates (θ,ø), and time (t). The entire globe is tessellated into geographical areas containing spectrum users, called user domains. Each user domain is recursively tessellated into smaller user domains, or subdomains. A computer cluster, or broker, in each smallest subdomain performs the calculations necessary to determine if a particular user in the subdomain experiences interference. Throughout this thesis, the Chicago Loop (area of 4.09 km2, population ~ 21, 000) is taken to be the representative example of a smallest subdomain.
Within each subdomain, the number of users served by a broker is reduced to a manageable number by the process of culling. There are three orders of culling. In first-order culling, subdomain pairs without line of sight and not close enough to mutually interfere are culled, or removed from further consideration for interference calculations. In second-order culling, within each subdomain an intersection test of the electrospace manifolds of all user pairs is performed. User pairs whose manifolds do not intersect are culled. In third-order culling a Friis calculation is performed for all remaining user pairs. The output of third-order culling is an RFI flag bit for each user indicating whether interference is present or not. The computational complexity of first-, second-, and third-order culling calculations was determined.
Three representative user classes will be discussed: WiFi access points, Terminal Doppler Weather Radars, and passive EESS satellites. The manifold descriptor language (MDL) for each of the three user classes was described. The computational complexity of broker calculations to determine electrospace parameters from the MDL was determined. Using this complexity and the complexity of culling calculations, the total computational requirements for a broker in a representative subdomain is determined in GFLOPS (Giga Floating Point Operations Per Second).</p
A 449 MHz Modular Wind Profiler Radar System
Full text linkThis thesis presents the design of a 449 MHz radar for wind profiling, with a focus on modularity, antenna sidelobe reduction, and solid-state transmitter design. It is one of the first wind profiler radars to use low-cost LDMOS power amplifiers combined with spaced antennas. The system is portable and designed for 2-3 month deployments. The transmitter power amplifier consists of multiple 1-kW peak power modules which feed 54 antenna elements arranged in a hexagonal array, scalable directly to 126 elements. The power amplifier is operated in pulsed mode with a 10% duty cycle at 63% drain efficiency. The antenna array is designed to have low sidelobes, confirmed by measurements. The radar was operated in Boulder, Colorado and Salt Lake City, Utah. Atmospheric wind vertical and horizontal components at altitudes between 200m and 4km were calculated from the collected atmospheric return signals.
Sidelobe reduction of the antenna array pattern is explored to reduce the effects of ground or sea clutter. Simulations are performed for various shapes of compact clutter fences for the 915-MHz beam-steering Doppler radar and the 449-MHz spaced antenna interferometric radar. It is shown that minimal low-cost hardware modifications to existing compact ground planes of 915-MHz beam-steering radar allow for reduction of sidelobes of up to 5dB. The results obtained on a single beam-steering array are extended to the 449 MHz triple hexagonal array spaced antenna interferometric radar. Cross-correlation, transmit beamwidth, and sidelobe levels are evaluated for various clutter fence configurations and array spacings. The resulting sidelobes are as much as 10 dB below those without a clutter fence and can be incorporated into existing and future 915 and 449 MHz wind profiler systems with minimal hardware modifications
Mathematical Formulation of the Remote Electric and Magnetic Emissions of the Lightning Dart Leader and Return Stroke
Full text linkLightning detection and geolocation networks have found widespread use by the utility, air traffic control and forestry industries as a means of locating strikes and predicting imminent recurrence. Accurate lightning geolocation requires detecting VLF radio emissions at multiple sites using a distributed sensor network with typical baselines exceeding 150 km, along with precision time of arrival estimation to triangulate the origin of a strike. The trend has been towards increasing network accuracy without increasing sensor density by incorporating precision GPS synchronized clocks and faster front-end signal processing. Because lightning radio waveforms evolve as they propagate over a finitely conducting earth, and that measurements for a given strike may have disparate propagation path lengths, accurate models are required to determine waveform fiducials for precise strike location. The transition between the leader phase and return stroke phase may offer such a fiducial and warrants quantitative modeling to improve strike location accuracy.
The VLF spectrum of the ubiquitous downward negative lightning strike is able to be modeled by the transfer of several Coulombs of negative charge from cloud to ground in a two-step process. The lightning stepped leader ionizes a plasma channel downward from the cloud at a velocity of approximately 0.05c, leaving a column of charge in its path. Upon connection with a streamer, the subsequent return stroke initiates at or near ground level and travels upward at an average but variable velocity of 0.3c. The return stroke neutralizes any negative charge along its path. Subsequent dart leader and return strokes often travel smoothly down the heated channel left by a preceding stroke, lacking the halting motion of the preceding initial stepped leader and initial return stroke. Existing lightning models often neglect the leader current and rely on approximations when solving for the return stroke.
In this Thesis, I present an analytic solution to Maxwells Equations for the lightning leader followed by a novel return stroke model. I model the leader as a downward propagating boxcar function of uniform charge density and constant velocity, and the subsequent return stroke is modeled as an upward propagating boxcar with a time dependent velocity. Charge conservation is applied to ensure self-consistency of the driving current and charge sources, and physical observations are used to support model development. The resulting transient electric and magnetic fields are presented at various distances and delay times and compared with measured waveforms and previously published models
A computational study of dielectric photonic-crystal-based accelerator cavities
Full text linkFuture particle accelerator cavities may use dielectric photonic crystals to reduce harmful wakefields and increase the accelerating electric field (or gradient). Reduced wakefields are predicted based on the bandgap property of some photonic crystals (i.e. frequency-selective reflection/ transmission). Larger accelerating gradients are predicted based on certain dielectrics’ strong resistance to electrical breakdown. Using computation, this thesis investigated a hybrid design of a 2D sapphire photonic crystal and traditional copper conducting cavity. The goals were to test the claim of reduced wakefields and, in general, judge the effectiveness of such structures as practical accelerating cavities. In the process, we discovered the following: (1) resonant cavities in truncated photonic crystals may confine radiation weakly compared to conducting cavities (depending on the level of truncation); however, confinement can be dramatically increased through optimizations that break lattice symmetry (but retain certain rotational symmetries); (2) photonic crystal cavities do not ideally reduce wakefields; using band structure calculations, we found that wakefields are increased by flat portions of the frequency dispersion (where the waves have vanishing group velocities).
A complete comparison was drawn between the proposed photonic crystal cavities and the copper cavities for the Compact Linear Collider (CLIC); CLIC is one of the candidates for a future high-energy electron-positron collider that will study in greater detail the physics learned at the Large Hadron Collider. We found that the photonic crystal cavity, when compared to the CLIC cavity: (1) can lower maximum surface magnetic fields on conductors (growing evidence suggests this limits accelerating gradients by inducing electrical breakdown); (2) shows increased transverse dipole wakefields but decreased longitudinal monopole wakefields; and (3) exhibits lower accelerating efficiencies (unless a large photonic crystal is used)
Calibration of millimeter-wave radiometers with application to clear-air remote sensing of the atmosphere
No full textPh.D
A 449 MHz Modular Wind Profiler Radar System
Full text linkThis thesis presents the design of a 449 MHz radar for wind profiling, with a focus on modularity, antenna sidelobe reduction, and solid-state transmitter design. It is one of the first wind profiler radars to use low-cost LDMOS power amplifiers combined with spaced antennas. The system is portable and designed for 2-3 month deployments. The transmitter power amplifier consists of multiple 1-kW peak power modules which feed 54 antenna elements arranged in a hexagonal array, scalable directly to 126 elements. The power amplifier is operated in pulsed mode with a 10% duty cycle at 63% drain efficiency. The antenna array is designed to have low sidelobes, confirmed by measurements. The radar was operated in Boulder, Colorado and Salt Lake City, Utah. Atmospheric wind vertical and horizontal components at altitudes between 200m and 4km were calculated from the collected atmospheric return signals.
Sidelobe reduction of the antenna array pattern is explored to reduce the effects of ground or sea clutter. Simulations are performed for various shapes of compact clutter fences for the 915-MHz beam-steering Doppler radar and the 449-MHz spaced antenna interferometric radar. It is shown that minimal low-cost hardware modifications to existing compact ground planes of 915-MHz beam-steering radar allow for reduction of sidelobes of up to 5dB. The results obtained on a single beam-steering array are extended to the 449 MHz triple hexagonal array spaced antenna interferometric radar. Cross-correlation, transmit beamwidth, and sidelobe levels are evaluated for various clutter fence configurations and array spacings. The resulting sidelobes are as much as 10 dB below those without a clutter fence and can be incorporated into existing and future 915 and 449 MHz wind profiler systems with minimal hardware modifications
- …
