In this dissertation, distributed control with sensor networks is used to suppress beam vibrations. The distributed active vibration control strategy has the advantages of being scalable and fault-tolerant for use in large-scale systems. A distributed control system normally consists of numerous localized controllers called nodes. Each localized controller has a sensor, an actuator and a means of communicating with other nodes in the system. The goal of distributed control is to achieve global performance by sharing sensor information among the localized controllers. This is in contrast to decentralized control whose localized controllers work independently to achieve a global performance, and centralized control which utilize one central processor. <p> A simply supported beam with six pairs of piezoelectric transducers acting as sensors and actuators is the active structure investigated. The disturbance on the beam is band-limited white noise (0 - 600 Hz). The dynamics of the beam are obtained using experimental system identification, and a 36 state model is selected for control design use after exploring various model sizes. Since existing distributed control design approaches are not applicable to structural vibration systems due to the strongly connected nature of vibration system dynamics, new distributed controllers are designed based on traditional H2 optimal control theory. Such H2 optimal control has been proven effective and robust at attenuating structural vibration in centralized strategy, and it is extended here to a distributed architecture. Two types of sensor grouping strategies in the distributed control system are considered: groups based on physical proximity and groups based on modal sensitivity. Distributed middleware services such as clock synchronization and network communications routing are also investigated and implemented experimentally for vibration control. This work is the first experiment implemented in the distributed vibration control field, and control performance results demonstrate the effectiveness of the two distributed grouping approaches. <p> A fault-tolerant active vibration control system is applied to a simply supported beam with high order. System failures are detected and isolated by Beard-Jones (BJ) filters, and then a controller specifically designed for the faulty system is switched on, in order to maintain optimal control performance and stability under failure conditions.

This thesis examines the suitability of the Verbs and Adverbs algorithm in the context of motion generation for humanoid robotic systems. The Verbs and Adverbs algorithm is a non-linear interpolation algorithm originally designed as a method for generating realistic motions in real-time for computer-animated figures. In this algorithm, Verbs represent a behavior and are parameterized by Adverbs. Motion examples are encoded as a matrix of linear and non-linear coefficients in adverb space. The suitability of the algorithm, in this context, is defined in terms of the ability to integrate this algorithm into a humanoid robotic system as well as the ability of the algorithm to rapidly generate new motions for playback on that platform.

The amount of charge required to represent a logic state in CMOS digital circuits has been reduced dramatically with the scaling of supply voltage and nodal capacitances, making radiation-induced single event effects increasingly problematic. Circuit hardening approaches, such as Triple Mode Redundancy (TMR) and Dual Interlocked Storage Cell (DICE latch) have been employed to address this issue; however many of these techniques are designed to mitigate effects of charge deposited at a single circuit node. Decreased spacing of devices with scaling can increase the charge collection at nodes other that than the hit node. Such charge collection at multiple nodes due to a single hit (i.e. charge sharing) can render existing methods for Single Event Upset (SEU) mitigation ineffective. Using three-dimensional (3D) device-level TCAD simulations, the mechanisms responsible for charge sharing were determined for the 130 nm technology node. Parasitic bipolar turn-on is shown to be the primary mechanism for charge sharing between two PMOS transistors, whereas charge diffusion is the primary mechanism responsible for charge sharing in NMOS transistors. Mitigation techniques explored showed that the use of a contacted guard-band reduced the charge sharing between PMOS devices in the n-well by 97% and between NMOS devices in the p-well by 35%. A combination of circuit simulations, 3D TCAD simulations, and mixed-mode simulations show that charge sharing between sensitive pairs of devices is the primary reason for experimentally observed upsets in the DICE latch design when exposed to low LET ions. The use of contacted guard-band, interdigitation, and nodal separation of sensitive nodes on the layout level can be employed to retain the hardness of the DICE latch and other circuit level hardening techniques.

CVD diamond has proven to be a viable material for creating electrical devices. This research focuses on the advanced development and characterization of CVD diamond dielectric films. It is shown in this research that the single most important parameter in the development of diamond dielectric capacitors is electrical conductivity. Conductivity affects all electric mechanisms in a polycrystalline diamond film. Part of this research effort was devoted to analyzing the conduction mechanisms of CVD diamond films. The control and/or suppression of these conduction mechanisms determine the dielectric properties of CVD diamond. This research also focused on the design and development of advanced capacitor structures using CVD diamond. Earlier published work on diamond dielectrics has been limited to basic characterization experiments with simple capacitive structures. Experiments proceeded in this research that produced multi-layer diamond capacitor designs and prototypes.

Preliminary experimentation shows diamond emits electrons at high temperature through a phenomenon known as thermionic emission. This thesis details an experiment to quantify the thermionic emission of a diamond on silicon vacuum cathode at high temperatures and to test the efficiency of thermal heating as a means of emission. Using a pulsed laser as a heat source for the cathode, experiments were run to determine the emission characteristics of both a diamond-coated silicon cathode and a bare silicon cathode. The results of each device are compared. Results of the experiment show a thermionic emission characteristic following a Schottky emission plot for each device. An Arrhenius plot was made from Richardsons equation to show the activation energy required for the diamond on silicon device to make a transition from silicon emission to diamond emission.

<p>There is an ever increasing concern about security threats as embedded systems are moving towards networked applications. Model based approaches have proven to be effective techniques for embedded systems design. However, existing modeling tools were not designed to meet the current and future security challenges of networked embedded systems. In this thesis, a framework to incorporate security modeling into embedded system design is presented. A security analysis tool is presented which analyzes information flows within an embedded system for conformance to the Bell-LaPadula and Biba security models. This tool can easily integrate with existing tool chains to create co-design environments that address security, functionality and system architecture aspects of embedded systems concurrently. A case study is presented where the security analysis tool is integrated with SMoLES, an embedded systems design language, to create one such co-design environment.</p>

Total ionizing dose effects in a low-dropout voltage regulator are explained based on experimental data and circuit simulations. Transistor gain degradation is shown to be the dominant cause of the circuit degradation at lower dose rates. In addition, collector-to-emitter leakage current in one of the NPN transistors of the bandgap reference part of the circuit is responsible for increasing the postirradiation output voltage at high dose rates. Parametric changes in the bandgap, differential amplifier, and output pass transistor circuit blocks are identified that are responsible for various aspects of the observed circuit degradation. The different annealing characteristics of oxide-trap and interface-trap charge are responsible for the complex postirradiation recovery of the output voltage.

Estimating egomotion from a video sequence is intrinsically difficult and requires high-level mathematics and programming skills. This work exploits existing technology to leverage the development of a mobile robots navigation. An open source software MPEG encoder package is modified so that its motion vectors and encoded frame type are accessible. As a result, the process of estimating a motion field from the MPEG motion vectors is far less complicated and time-consuming than those used in conventional methods. The main contribution is the creation of low-cost multiple sensors for a mobile robot. Two real-time applications, visual odometry and precipice detection, are presented. Despite employing simple trigonometry, the visual odometry performs consistently well on moderately textured surfaces with low specular reflection. A proposed novel approach to detecting a precipice in real-time is shown to be successful even when the robot runs at a very high speed. The experimental results substantiate the use of both applications in real situations.

Robot-assisted rehabilitation has been an active research area for the last few years to automate therapy for regaining mobility with arm and hand movements following a deficiency in facility due to stroke. However, task-oriented therapy approaches that require patients to practice complex and more-functional activities of daily living (ADL) tasks cannot be performed by the existing robot-assisted rehabilitation systems because they are only limited to providing assistance to either arm or hand movement. Therefore, an intelligent controller for robot-assisted rehabilitation systems is desirable in order to perform ADL tasks that generally require coordination of both arm and hand movements. In this dissertation, an intelligent control architecture is designed to coordinate in-house designed assistive devices in a systematic manner to enable the stroke patients to perform ADL tasks. The proposed control architecture is the first of its kind that brings the benefit of coordination of arm and hand assistive devices which is expected to address the deficit of coordinated assistive devices in the field of rehabilitation robotics. The control architecture is represented in terms of a hybrid system model combining a high-level controller for decision-making and two low-level assistive controllers (arm and hand controllers) for providing arm and hand motion assistance. The application of a hybrid system model for rehabilitation purposes is unique. Furthermore, providing robotic assistance to the patients to complete the rehabilitation task in a smooth manner is an important objective in rehabilitation therapies. Thus, the low-level assistive controllers in the control architecture are designed in such a way as to enhance smooth human-robot interaction involving the subject and the robotic assistive devices. Results from real-time assistance experiments on unimpaired subjects are presented to demonstrate the efficacy of the presented control architecture.

Trustworthy computer systems protect the access of sensitive information by an unauthorized agent (i.e., an attacker). However, security vulnerabilities exist which can allow a system to be compromised by an attacker. An attacker could simply download physical memory to recover sensitive information. Encrypting memory utilized by the processor can provide protection against this type of access, but inserting an encryption module between the processor and memory has its challenges. Proposed here is a method for viably encrypting the data channel using an FPGA between processor and memory in an effort to minimize significant redesigns for the processor. First, an overview of security on FPGAs is discussed, highlighting advantages and problems. Second, a discussion of an implementation of the DES algorithm on an FPGA is covered in depth. Next, an overview of the entire system, processor, encryption and memory, is detailed. Finally, it concludes with a discussion of the performance experiment and analysis.