<p> For an engineer, it is always a difficult task to deploy a distributed wireless sensor network for buildings, without initial planning of the infrastructure of the network. A robust network is one that provides entire connectivity without compromising the signal strength. Another important parameter to be considered is the cost of the network, which can be reduced by deploying the most optimized and resilient network. With this project, we have designed an interactive design tool using Matlab that can help an engineer to plan and place the router nodes before actual deployment. The reference methods are based on Dijkstra&rsquo;s algorithm, line based algorithm and clustering algorithm for placing of routers. In this project, we have used best from all the algorithms to present a novel approach for the placement of routers. This algorithm provides improved results than the reference methods and thus further reduces the cost of the network.</p>

<p> The use of hand gesture recognition has been steadily growing in various human-computer interaction applications. Under realistic operating conditions, it has been shown that hand gesture recognition systems exhibit recognition rate limitations when using a single sensor. Two dual-sensor approaches have thus been developed in this dissertation in order to improve the performance of hand gesture recognition under realistic operating conditions. The first approach involves the use of image pairs from a stereo camera setup by merging the image information from the left and right camera, while the second approach involves the use of a Kinect depth camera and an inertial sensor by fusing differing modality data within the framework of a hidden Markov model. The emphasis of this dissertation has been on system building and practical deployment. More specifically, the major contributions of the dissertation are: (a) improvement of hand gestures recognition rates when using a pair of images from a stereo camera compared to when using a single image by fusing the information from the left and right images in a complementary manner, and (b) improvement of hand gestures recognition rates when using a dual-modality sensor setup consisting of a Kinect depth camera and an inertial body sensor compared to the situations when each sensor is used individually on its own. Experimental results obtained indicate that the developed approaches generate higher recognition rates in different backgrounds and lighting conditions compared to the situations when an individual sensor is used. Both approaches are designed such that the entire recognition system runs in real-time on PC platform.</p>

Present trends of increasing health care costs, demand novel methods and tools for a transformation from a reactive and hospital-driven to a proactive, patient-centered and enabling healthcare. This work presents a modular wireless healthcare platform, that can be customized to meet different clinical and patient needs. It offers multiple sensing modalities to acquire a variety of physiological signals including bio-potentials, thoracic impedance, activity and posture, blood pressure, pulse wave form and blood oxygen saturation. A customized firmware allows for real-time digital signal processing to extract features, immediate alarm, feedback, and/or interface to therapeutic devices and procedures. The integration of our platform into mobile device technologies provides control and visualization of patient health data during treatment. Thus, patients can become actively engaged in their health and take responsibility leading to a potentially higher adherence to the treatment plan and early diagnosis. Real-time monitoring and immediate data transfer to a secured cloud based database provides decision support for physicians and caregivers and enables data mining to identify early markers for timely intervention in order to reduce costly hospital stays or re-admissions.

The increasing demand for local high-frequency operations on microprocessor and data-communication chips has led to the need for phase-locked loops to generate on-chip high-frequency clocks controlled by a much lower-frequency externally-provided system reference clocks. A high-frequency integrated CMOS phase-locked loop, which is independent of temperature and silicon processing variations, has been designed and tested. The system-level design was based on both frequency linear analysis and transient time-domain analysis performed using MTIME simulations. The design in this work integrates a phase-frequency detector, a symmetric charge-pump, an external low-pass RC loop filter, a digital-load-controlled voltage-controlled oscillator (DLCVCO), a programmable feedback-divider, a frequency-range detector (FRD), a circuit which initializes the loop filter node to VDD, a digital-load-activation cell, and associated circuitry on a portion of one die. The DLCVCO was implemented by a ring of three inverters; frequency was controlled by digitally programming the capacitive loads connected to its inner nodes. The FRD was designed to sense the temperature and silicon process conditions of the PLL system, and control the number of capacitive loads activated to the inner nodes of the DLCVCO. A 0.8 μm CMOS process was used to implement this PLL system design. Tests demonstrated that this PLL generated frequencies in excess of 200MHz, locked onto a 2MHz reference clock, and achieved a tuning range of 12MHz to 212MHz independent of silicon process variations and temperature conditions.

Suicide is a major health problem in the US, and has become an important topic of study. Recently, psychiatrists have reported hearing a particular sound or tonality in the voices of subjects that are at high risk of attempting suicide. This has lead to research to discriminate between depression and this high risk state based on the acoustic properties of the subject's voice, and has lead to some very promising results. Many of the features that have been used for this task are based on the power spectrum of the voice; however, it is not clear whether these features have captured the particular tonality that the psychiatrists have reported. The whole work attempts to address this question through analysis of speech data by calculating the harmonics, amplitudes, and other discriminating features from 6 depressed and 6 high risk subjects.

Robust localization is essential for successful navigation of autonomous mobile robots in any environment. Advanced techniques such as simultaneous localization and mapping (SLAM) also consider reliable localization as a critical component. The major existing localization algorithms like Particle Filtering, however, are, based on the assumption that the environment is static which does not hold true in some practical applications of mobile robots. In this thesis, we explore a rather fledgling area of wireless robotic capsule endoscopy where dynamism is an inherent feature of the environment. Since the last decade, intense research has spurred in the field of capsule robots used for inspecting the entire gastrointestinal (GI) tract for various diseases, such as obscure gastrointestinal bleeding (OGIB), tumors, cancer, Crohns disease, and celiac disease. It is envisioned that the next generation of these micro-robots will traverse the GI tract with greater autonomy. In such self-propelling tiny robots, the incorporation of mini-surgical tools and artificial intelligence is anticipated. Capsule robot localization and mapping of the environment inside the human body will be a key essential for successful autonomous navigation.

The terminal impedance of a full-wave dipole as a function of centre spacing has been measured for dipoles with the ratio of wave length to diameter (λ/D) equal to : 188, 94.3, 75.7, 62.9, 47.2, and 23.6 for centre spacings (S/λ)of .0212, .0423, .074, and .105. All measurements were made at 1000 megacycles using a standing wave measuring method with the half dipole above an image plane. The variation of resistance and reactance of the terminal impedance with centre spacing is shown in a series of graphs. A considerable increase in anti-resonant resistance and in antenna length for anti-resonance is observed to occur with increased gap length. The results are shown to agree well with existing literature.

The response of electronic devices to ionizing radiation is a reliability concern for commercial and space applications. An ionizing particle can cause charge to be generated and collected at the node of a circuit. If the charge is collected in the memory element, a single event upset (SEU) can occur. Since the discovery of single event effects, circuit designers have faced the challenge of creating radiation hardened by design (RHBD) flip-flops that decrease the memory elements susceptibility to SEUs. Test structures of unhardened and hardened flip-flop designs are fabricated and irradiated in order to understand the relative error rates due to single event upsets. Each test structure is irradiated using heavy-ion particle accelerators to simulate the radiation in a space environment. In the past only particles with a relatively high linear energy transfer (LET) were capable of generating enough charge to flip the state of a memory element. However, as technology has scaled, the critical charge necessary to cause an SEU has decreased. Radiation effects must now be considered in calculating the error rate for commercial applications as well. Irradiating circuits using isotropic button sources simulates terrestrial radiation such as alpha particles emitted from packaging material. The continued improvement of RHBD techniques relies on the ability to thoroughly test circuits at each new technology node over all radiation environments and understand how the cells upset. In this thesis, SEU data for hardened and unhardened flip-flop designs from a variety of radiation sources are presented. The combination of data from heavy-ion beam, alpha particle beam, and isotropic alpha particle sources provides the critical parameters for calculating error rate for a given flip-flop design in a specific environment. Heavy-ion data also reveals the robust design of DICE, DICE Guard Gate, and Stacked flip-flops in comparison with unhardened flip-flop designs in 32 nm SOI technology.

<p> Field oriented control (FOC) and direct torque control (DTC) are the two most important control strategies for modern advanced machine drives. Field oriented control (FOC) has been widely used as the industrial motor drives. The direct torque control has been recently developed by Takashi and has shown great potential in machine drive applications. This thesis discusses the performances of direct torque control on AC machine drives including the industrial workhorse induction machines (IM) and permanent magnet synchronous machines (PMSM). Both of the IM and PMSM dynamics models are derived in details in the thesis. By applying nonlinear filtering techniques, the sensorless direct torque control of AC machine drives are studied. MATLAB simulations have shown the efficacy and superior performance of the resilient extended Kalman filter over the traditional extended Kalman filter for direct torque control applications.</p>

Traditional upper extremity prostheses are comprised of an elbow joint and a terminal device (often either a split-hook or a hand) and have typically been limited to a single control input. In the case of a myoelectric prosthesis, a single pair of electromyograms (EMG) inputs from the biceps and triceps muscles, respectively, is typically used to control both the elbow and terminal device. Dependence on a single input forces the prosthesis to be controlled in a sequential manner, where either the hand or elbow is controlled at any given time. With fewer inputs to control more joints, new control strategies are required to provide a transhumeral amputee with control of a limb that feels intuitive and natural. The objective of this work is the development of highly functional and intuitive coordinated controller for an arm prosthesis that exhibits improved functionality relative to existing prostheses. This work leverages recent advances in MEMS in order develop a controller that enables the simultaneous movement of the anatomical shoulder joint and prosthetic elbow joint, and also enables switching between control of a hand and elbow, and is used in conjunction with EMG input to provide enhanced whole arm (hand and elbow) control, relative to the traditional conventional control approach. This work involves implementing the control methods on a prosthesis prototype and conducting studies on amputee subject to assess the effectiveness of the coordinated controller in performing different reaching tasks relative to traditional sequential control devices.