Presently the tools available for predicting the effect of radiation on microelectronic circuits rely on SPICE models to simulate post-irradiation parameter and performance changes. However, modeling complex integrated circuits with SPICE simulations requires large amounts of engineering and computer simulation time. Also, repetitive simulations are needed with SPICE to simulate the evolution of degradation in the circuit parameters with total dose. Behavioral modeling technique provides an excellent alternative. Simplified behavioral models of a circuit or a sub-circuit could replace the SPICE-based circuits. The behavioral models describe the electrical circuit behavior and its dependence on the radiation dose with a high degree of accuracy. A single continuous simulation could ideally cover the entire circuit response from normal electrical operation (pre-irradiation or pre-rad) to post-irradiation performance at different exposure levels to assess system failure or to qualify radiation tolerance. This thesis describes a generic modeling technique to create TID aware behavioral models of voltage feedback op-amps without creation of underlying SPICE micro-model. The op amp behaviors dependence on supply voltage, input voltage and total dose are captured in the model. The model accurately predicts TID response of board-level designs without expensive fabrication runs. The behavioral model runs at least 79 times faster than SPICE in the case of Schmitt trigger oscillator circuit while maintaining accuracy to within 5% of SPICE values.

Size and complexity of software systems are increasing and there is increasing demand for component based distributed applications and systems. Performance characteristics such as throughput and scalability are crucial quality attributes of such systems. For this reason, it is very critical to validate that the system satisfies the performance requirements. Performance testing is a way to evaluate performance characteristics. Large scale distributed systems often have stringent performance requirements. Throughput, latency, scalability are important performance metrics for such systems. For this reason, performance testing plays an important role in middleware based distributed systems. In this thesis, it will be recognized that there is a need for a way to characterize and capture performance characteristics of components and the component model in a distributed system so that the effect of complex component interactions on system performance can be explored. In order to be able to test the performance of the system by taking into account the couplings of components and middleware, component interactions should precisely be understood and captured from a performance perspective in a component oriented performance model. An approach and an implementation focusing on this need will be presented in this thesis.

This work describes a control architecture and intent recognition approach for the real-time supervisory control of a powered lower limb prosthesis. The proposed approach infers user intent to stand, sit, or walk, by recognizing patterns in prosthesis sensor data in real-time, without the need for instrumentation of the sound-side leg. Specifically, the intent recognizer utilizes time-based features extracted from frames of prosthesis signals, which are subsequently reduced to a lower dimensionality (for computational efficiency). These data are initially used to train intent models, which classify the patterns as standing, sitting, or walking. The trained models are subsequently used to infer the users intent in real-time. The effectiveness of the proposed approach is demonstrated via experiments with a single unilateral amputee subject. Additionally, design of real-time slope and cadence estimators using sensors on the prosthesis is described. The extracted slope and cadence information is used to switch between different sets of controller parameters for walking on different slopes and speeds.

The 1/f noise magnitude of n-channel metal-oxide-semiconductor field effect transistors is found to decrease by up a factor of ~3 after 18 years of room-temperature aging. This decrease is largest in devices with high-temperature post-gate-oxidation nitrogen annealing, which increases the densities of O vacancies and strained Si-Si bonds near the silicon/silicon dioxide interface. Approximately 100 mV positive threshold voltage shifts are observed for all device types during aging; these are attributed to reactions with hydrogenous species (e.g., water vapor). The aging related changes in 1/f noise may well be caused by the relaxation of strained Si-Si bonds associated with O vacancies in the near-interfacial silicon dioxide layer.

Defects that lie at or near the semiconductor-oxide interface of MOS transistors were characterized using 1/f noise and charge pumping measurements. The frequency, gate-voltage, and temperature dependence of the noise were investigated for moisture-exposed and control Si nMOS and pMOS transistors before and after irradiation. Moisture exposure did not significantly change nMOS device noise after irradiation compared to control devices. For the exposed pMOS parts, significant changes in the noise with irradiation occurred, which reflect changes in the trap density and energy distributions. Three-level charge pumping measurements, in conjunction with noise measurements, were used to probe the trap densities of these pMOS devices before and after irradiation. For the exposed device, enhanced radiation-induced border trap densities were observed through an increase in the charge recombined charge per cycle, consistent with the increase in noise and the change in gate-voltage dependence of the noise. The techniques described above were applied to Ge pMOS transistors to gain insight into the properties of the semiconductor/gate-dielectric interface, and to determine the effects of processing on the defects within that region. The number of silicon monolayers and the halo implantation dose strongly affect the radiation response and 1/f noise levels of these devices. In addition, significantly different border trap and interface trap energy distributions were estimated via 1/f noise and charge pumping measurements, strongly suggesting that the noise in these devices is dominated by bulk oxide traps in the dielectric layer.

This dissertation addresses the problem of how to make technology-based intervention tools for children with autism spectrum disorders (ASD) affect-sensitive. Two computer-based cognitive tasks are designed to elicit the affective states of liking, anxiety, and engagement. A large set of physiological indices are investigated. Subjective reports on affective states from a therapist, a parent, and the child are collected and analyzed. Therapist-like affective models are designed using Support Vector Machines, which yields 82.9% prediction. The models are applied during affect-sensitive human-robot interaction where the robot adapts its behaviors to detected affective states of a child with ASD. This success motivates the design and evaluation of realistic social interaction scenarios for ASD children using virtual environments. The results support the viability of physiology-based affective computing for future ASD intervention.

Detecting malicious modifications to a circuit is a daunting task, regardless of the medium. In a fabricated circuit, most methods of detecting hardware Trojans rely on small changes in side-channel measurements, which can easily be disturbed by the presence of severe process variation. In a Field Programmable Gate Array (FPGA), the reprogrammability and design transparency reduces the initial difficulty of inserting a Trojan to a circuit. This thesis suggests solutions for both situations. The first method uses a signature generated from altering the supply voltage to the circuit in a controlled manner; this process leads to a change in shape of the transient current response of the integrated circuit (IC). Simulation results presented show significant differences between circuits with and without malicious hardware (i.e., Trojans), despite large variations in individual transistor parameters. Second is a Trojan-protection by design method where a controllable ring-oscillator is inserted into a circuit in order to detect modifications that change the timing of its component gates. This design was able to win 2nd place in a competition held by the Polytechnic Institute of New York University.

This thesis presents the framework and implementation of a mobile air quality monitoring network, with an in-depth discussion of several new innovative techniques for web-based visualization. These techniques allow typical web users to access high-resolution pollution data gathered from a large number of vehicle-mounted mobile sensing devices coupled with highly-accurate static sensor data in an easy-to-use, intuitive interface. Additionally, this interface offers users a set of novel applications to promote health and pollution awareness, including a green trip planner, whereby users can plot routes between two locations based on a path of least exposure to specied pollutants, and an exposure estimator, which allows users to calculate previous levels of exposure to harmful pollutants based only on a single timed GPS track. The work described in this paper includes research on hardware aspects of the system pertaining to the mobile sensing nodes, communications structure and protocols, as well as software aspects pertaining to the implementation of the aforementioned flash-based web client.

Soft errors can alter the correct execution of code within a microprocessor, particularly if control logic is compromised. This thesis addresses the vulnerability of a microprocessors control logic by assigning a signature to each instruction; this signature is based upon the expected length of time it takes for the instruction to retire. This information is then compared to the actual retirement time and generates an error signal when a mismatch occurs. A VHDL description of a MIPS R2000 processor has been modified to test this concept. The processor was implemented in a Field Programmable Gate Array (FPGA) for fault injection and simulation using the Dhrystone benchmark. In addition, it was synthesized with the Oklahoma State University FreePDK 45 nm System on Chip Library for timing and area comparative analysis. The results showed that our proposed design reduces the control logic's vulnerability to soft errors by more than 80% while adding less than 1% overhead.

Direct charge collection measurements are presented which prove that the presence of tungsten near sensitive volumes leads to extreme charge collection events through nuclear reactions. We demonstrate that, for a fixed incident particle linear energy transfer (LET), increasing particle energy beyond a certain point causes a decrease in nuclear reaction-induced charge collection. This suggests that a worst-case energy exists for single-event effect susceptibility, which depends on the technology, device layout, and the incident ions fixed LET value. A Monte Carlo approach for identifying the worst-case energy is applied to certain bulk-Si and silicon-on-insulator technologies. Simulation results suggest that the decrease in charge collection beyond the worst-case energy occurs because the secondary particles produced from the high-energy nuclear reactions have less mass and higher energy and are therefore less ionizing than those produced by lower-energy reactions.