Plume location and prediction using mobile sensors is the main contribution of this thesis. Plume concentration values measured by chemical sensors at different locations are used to estimate the source of the plume. This is achieved by employing a stochastic approximation technique to localize the source and compare its performance to the nonlinear least squares method. The source location is then used as the initial estimate for the boundary tracking problem. Sensor measurements are used to estimate the parameters and the states of the state space model of the dynamics of the plume boundary. The predicted locations are the reference inputs for the LQR controller. Measurements at the new locations (after the correction of the prediction error) are added to the set of data to refine the next prediction process. Simulations are performed to demonstrate the viability of the methods developed. Finally, interpolation using the sensors locations is used to approximate the boundary shape.

Positron Emission Tomography (PET) has been a very useful laboratory tool for the noninvasive study of dynamic physiological processes within the human body. Its clinical worth in cardiology, neurology, and oncology has been verified for well over a decade. Only with the most recent introduction of high performance analog and VLSI digital components yielding higher scanner resolutions at reduced costs, has clinical PET truly become a reality. The High Performance Detector Electronics System presented here represents indubitably the most advanced processing system available in the clinical PET market.

In this work the design of a constant-fraction discriminator (CFD) fabricated in the Orbit Semiconductor l.2-Jl n-well CMOS process is presented. This timing pick-off circuit is designed for use in the readout electronics of the Lead-Scintillator subsystem of the Pioneering High Eenergy Nuclear Ion eXperiment (PHENIX) Electromagnetic Calorimeter at the Relativistic Heavy Ion Collider (RHIC). The design was driven by stringent requirements including low power consumption, small area, arrayable, low cost and a fully integratable shaping network. Various integratable CFD shaping methods are investigated, and the candidate methods chosen for fabrication were the distributed R-C delay-line shaping, lumped-element R-C shaping and Nowlin method shaping. An additional channel of ideal delay-line shaping, utilizing coaxial cable to generate delay, was fabricated and used for a reference in comparing methods. These shaping methods are compared on the basis of die area, time walk performance and timing jitter performance as implemented using the CMOS CFD presented. Each shaping method investigated required no power from the dc supply. Die area for the distributed R-C delay-line, lumped-element R-C, Nowlin method and ideal delay-line (fraction circuit only) were 172 Jl X 70 Jl, 160 Jl X 65 Jl, 179 Jl X 53 Jl and 67 Jl X65Jl,respectively. Timewalkovera100:1dynamicrange(-2Vpeakto-20mVpeak) for these shaping methods in turn was found to be ± 175 ps, ± 150ps , ± 150 ps and ± 185 ps, respectively. Timing jitter performance with a minimum input signal (-20 mVpeak) in rms units for the four methods in turn were 65 ps, 85 ps, 100 ps and 65 ps. The average power dissipated per CFD channel was found to be approximately 12 mW.

The aerodynamic applications of plasma science is a field of growing interest. Investigations using various approaches have been initiated by several research groups that are designed to manipulate the aerodynamic boundary layer and to re-attach the flow to airfoils. EHD (ElectroHydroDynamic) flow control has proven at least as effective as other methods of boundary layer flow control. In the EHD approach, glow discharge plasma actuators are placed on the wings and fuselage of the aircraft, or on the turbine blades in the engine, to influence the boundary layer flow. This thesis is concerned with plasma actuators based on the OAUGDP ® (One Atmosphere Uniform Glow Discharge Plasma). An actuator consists of two conducting electrodes separated by a dielectric plate. When a sufficiently high RF voltage is supplied to the electrodes, the surrounding air ionizes and forms plasma in regions in which the electric field is above approximately 10 kV/cm. The ionized air, in the presence of an electric field gradient, produces a body force on the neutral gas flow.This work is concerned with two EHD effects: paraelectric flow acceleration and peristaltic flow acceleration. In the paraelectric mode, electric field gradients act on the net charge density of plasma, and the plasma drags the neutral gas along with it due to ion-neutral and electron-neutral Lorentzian collisions. In the peristaltic mode, successive actuators are energized with the same voltage, but increasing phase angles. The first part of this thesis describes experiments at the NASA Langley Research Center, Hampton, VA in the 7 x 11 Inch Low Speed Wind Tunnel in which Pitot tube velocity profile measurements and smoke flow visualization tests were conducted.The second part of this thesis describes the development of a low cost power supply to energize OAUGDP ® plasma actuators. The power supply consists of automotive ignition coil transformers, audio amplifiers, and a DC battery. Using this power supply, plasma actuators were energized at voltages up to 8 kV, and at frequencies between 0.5 and 8 kHz. This thesis also presents illustrative paraelectric flow acceleration data obtained using the low-cost power supply.

Positron Emission Tomography (PET) has been a very useful laboratory tool for the noninvasive study of dynamic physiological processes within the human body. Its clinical worth in cardiology, neurology, and oncology has been verified for well over a decade. Only with the most recent introduction of high performance analog and VLSI digital components yielding higher scanner resolutions at reduced costs, has clinical PET truly become a reality. The High Performance Detector Electronics System presented here represents indubitably the most advanced processing system available in the clinical PET market.

In this work the design of a constant-fraction discriminator (CFD) fabricated in the Orbit Semiconductor l.2-Jl n-well CMOS process is presented. This timing pick-off circuit is designed for use in the readout electronics of the Lead-Scintillator subsystem of the Pioneering High Eenergy Nuclear Ion eXperiment (PHENIX) Electromagnetic Calorimeter at the Relativistic Heavy Ion Collider (RHIC). The design was driven by stringent requirements including low power consumption, small area, arrayable, low cost and a fully integratable shaping network. Various integratable CFD shaping methods are investigated, and the candidate methods chosen for fabrication were the distributed R-C delay-line shaping, lumped-element R-C shaping and Nowlin method shaping. An additional channel of ideal delay-line shaping, utilizing coaxial cable to generate delay, was fabricated and used for a reference in comparing methods. These shaping methods are compared on the basis of die area, time walk performance and timing jitter performance as implemented using the CMOS CFD presented. Each shaping method investigated required no power from the dc supply. Die area for the distributed R-C delay-line, lumped-element R-C, Nowlin method and ideal delay-line (fraction circuit only) were 172 Jl X 70 Jl, 160 Jl X 65 Jl, 179 Jl X 53 Jl and 67 Jl X65Jl,respectively. Timewalkovera100:1dynamicrange(-2Vpeakto-20mVpeak) for these shaping methods in turn was found to be ± 175 ps, ± 150ps , ± 150 ps and ± 185 ps, respectively. Timing jitter performance with a minimum input signal (-20 mVpeak) in rms units for the four methods in turn were 65 ps, 85 ps, 100 ps and 65 ps. The average power dissipated per CFD channel was found to be approximately 12 mW.

Plume location and prediction using mobile sensors is the main contribution of this thesis. Plume concentration values measured by chemical sensors at different locations are used to estimate the source of the plume. This is achieved by employing a stochastic approximation technique to localize the source and compare its performance to the nonlinear least squares method. The source location is then used as the initial estimate for the boundary tracking problem. Sensor measurements are used to estimate the parameters and the states of the state space model of the dynamics of the plume boundary. The predicted locations are the reference inputs for the LQR controller. Measurements at the new locations (after the correction of the prediction error) are added to the set of data to refine the next prediction process. Simulations are performed to demonstrate the viability of the methods developed. Finally, interpolation using the sensors locations is used to approximate the boundary shape.

Memory design plays an important role in modern computer technology in regard to overall performance and reliability. Prior memory technologies, including magneticcore memory, hard disk drives, DRAM, SRAM have limitations in regard to bit density, IC integration, power efficiency, and physical size, respectively. To address these limitations we propose to develop a magnetic graphene random access memory (MGRAM) utilizing graphene Hall effect, which takes advantage of the inherent reliability of magnetic memory and superior electrical properties of graphene (high carrier mobility, zero-band gap, high Hall sensitivity). As the graphene magnetic memory device will be integrated with a CMOS ASIC design an analog circuit model for the MGRAM cell is necessary and important. In this study the electrical circuit model is developed utilizing the analog circuit modeling language Verilog-A. The electrical circuit model characterizes the graphene electrical properties and the ferromagnetic core magnetic properties that retains the bit-state value. MGRAM device simulations studying varying coil width, height, radius, contact pad configuration, graphene shape, is performed with the MagOasis Magsimus tool to evaluate the device performance. Model results show a maximum Hall effect voltage of 100mV for a bias current of 50uA with a 1 Tesla magnetic field, and a writing speed of 6-9ns for setting the magnetic state. These results will be validated against the circuit hardware measurement and will be used for model refinement.

Ground Penetrating Radar (GPR) is widely used in area of geologic exploration, hidden stationary subject detection and quality inspection on various infrastructures. The pulse generator, which offers very accurate timing information, is the most cardinal component in GPR systems. It is easy to design a pulse generator which produces pulse with pre-settled peak value and pulse width in nanosecond scale. However, since the system is working in complicated environments, various pulses in different pulse widths and amplitudes are needed. In this background, pulse generators in tunability and stability are precious in value and universal in use. Indeed, a few adaptive high voltage pulse signal generators in UWB circuit level have been developed. A pulse generator with tunable pulse width and controllable voltage amplitude is proposed under these demands. The proposed circuit implementation combines System-On-Chip (SOC) design with Printed Circuit Board (PCB) design because we intend to realize modulation separately. We also design an easy input console named Binary Input Array in the research to realize control simplicity. Furthermore, we employ mathematical model to optimize parameters in each component in order to have an improved performance. Simulation data are obtained from Cadence Virtuoso and OrCAD Capture.

The objective of this work is to design, simulate and synthesize a dynamic space controller system. The concept of space allocation and management can be applied to more than physical space. It may also be taken in the contexts of memory, network or bus management. The management and allocation of any space depends mostly on the twin factors of demand and availability. Time, the size and amount of space available, as well as the number of requests for the spaces, must also be considered. The proposed design has the capability to monitor multiple spaces for vacancies, length of time occupied and to dynamically track the number and location of available spaces. This system is easily programmed to adapt to the user’s application and is modular in design in order to facilitate expansion. The specific problem addressed in this work is a parking lot controller.