Search

In real-world conversational groups, interactants adjust their body position and orientation relative to one another in order to see and hear clearly. We use an agent-based modelling approach to compare alternative models for simulating the spatial patterns of conversational groups. The models are based on simple rules that control the movement, positioning, and orientation behaviour of individual agents, which in turn leads to the emergence of agent clusters. We identify which model alternative produces agent clusters with characteristics typical of real-world conversational groups. The centroid-based approach, where agents readjust their position and orientation with respect to the group centroid point, is a commonly used method to simulate conversational groups, but has not been empirically validated. This thesis replicates, evaluates, and validates the centroid-based model in a systematic way. Another model, where agents perform positional-orientational readjustments to see as many neighbours as possible within a 180 field of view, called the field-of-view approach is proposed, implemented, evaluated, and validated. Analysis of the spatial patterns of conversational groups has hitherto mostly relied on visual verification. We, novelly, use a combination of qualitative and quantitative methods to analyse the spatial patterns of conversational groups. Evaluations show that the field of- view model and centroid-based model produce agent clusters with significantly different social, spatial, and temporal characteristics. Validation is performed using a dataset which captures the spatial behaviour of 21 participants for the entire duration of a party. This validation shows that the characteristics of agent clusters resulting from the field-of-view model most closely reflects the characteristics of real-world conversational groups. We also show that a local neighbourhood influence works better than an extended neighbourhood influence to simulate conversational groups. The influence of objects in the environment on the spatial patterns of agent clusters are also discussed.

The use of adhesive bonding as a structural joining method has been gaining recognition in marine industry in recent years, though it has been widely adopted in other fields such as aerospace, automobiles, trains and in civil constructions. The type of materials used and design practices followed in marine structures are different from what is applied in other disciplines. Therefore new research approaches are required and recent novel ideas are ex- plored in the context of application of bonded joint configurations in marine environment. The research is directed at developing analysis tools for predicting the displacement, stress and strain fields in adhesively bonded joints between dissimilar adherends. In the finite element formulation, the adherends may be isotropic or orthotropic layered materi- als, which are assumed to behave linear elastically. The adhesive material is assumed to behave as elasto-plastic continuum, where the nonlinear behaviour is modelled as either a rigid or a semi-rigid adhesive solid that can be represented by the Ramberg-Osgood ma- terial model. The yield behaviour of the polymeric adhesive is modelled using a modified von Mises criterion, which accounts for the fact that plastic yielding of polymer materials may occur under the action of hydrostatic as well as deviatoric stresses. The geometric nonlinearity is based on the assumption of large displacement, large rotation but small strain, and it is implemented in the code using the total Lagrangian approach. The scheme is applied on three case studies viz.: a study of adherend imbalances in a single lap joint, stress analysis of a butt-strap joint system and a hybrid joint are un- dertaken. The influence of geometric and material nonlinearity on joint deformations and adhesive stresses, are studied for a single lap joint with dissimilar adherends, aluminium and a Fibre Reinforced plastic composite material, with varying adhrend thickness ratios. The adhesive stress-strain data obtained from the model are compared with the exper- imental stress-strain curve and the numerical results are validated with the analytical solution. Three dimensional effects like ’anticlastic’ and bending-twisting’ are shown in the joint with a dissimilar adherends. Key results are obtained that explains the state of nonlinear adhesive stress state in the joint. Analysis of butt-strap joint focussed on nonlinear modelling of a semi-rigid adhesive ma- terial that is used to bond two dissimilar adherends, steel and aluminium. The analysis demonstrate that the influence of geometric and material nonlinearity on the joint de- formations as well as the adhesive stresses is significant. Nonlinear adhesive stresses are compared with the actual strength of the highly flexible adhesive, highlighting the need for the consideration of material nonlinearity in the bonded joints. Failure modes for the joint are inferred from the observations made on the adhesive stress state in the butt-strap joint. Last study, deals with three dimensional analysis of a GRP-Steel hybrid joint carried out to model the initiation and propagation of crack under a set of static loading cases. Earlier studies were restricted only to two dimensional analysis. This three dimensional analysis showed that the adhesive normal stress is not constant across the width of the joint. Critical locations of stress concentrations are identified and the failure mechanisms are compared with the experimental specimens. The observations made from this research study using a three dimensional finite element program, compliments the present knowledge in the field of adhesively bonded joints.

Increasing reliance on wireless communication and complexity of cyberattacks have rendered industrial control systems (ICSs) such as process control systems (PCSs) (which are ICSs that operate chemical manufacturing processes) vulnerable to cyberattacks by malicious agents. In the past decade, several highly sophisticated cyberattacks (e.g., Stuxnet virus (2010), German steel mill attack (2014), Ukrainian power grid attack (2015), TRITON (2017)) have demonstrated that information technology (IT) infrastructure-based solutions to handling cyberattacks on control systems are insufficient on their own. An increasing body of research has focused on developing operational technology (OT)-based approaches to enhance the cyberattack resilience of PCSs. Cyberattack resilience here is defined as the ability of a PCS to minimize the impact of a cyberattack and recover from it. Research on cyberattack resilience of PCSs involves approaches that range from designing PCSs that are inherently attack-resilient to developing cyberattack detection, identification and mitigation schemes. Cyberattack detection schemes are OT-based anomaly detection schemes that reveal the presence of a cyberattack on a PCS by monitoring the process operational data for anomalies and are an important component of a cyberattack resilient PCS. The motivating realization behind the work presented in this dissertation is that the influence of PCS design parameters may be exploited to reveal the presence of an ongoing cyberattack on a PCS. In the chapters that follow, several approaches for cyberattack detection are presented. First, a control screening approach that may be used to incorporate attack detectability within the conventional PCS design considerations is presented. The screening algorithm is based on a characterization of the interdependence between the PCS design parameters, and the ability of the detection scheme to detect the attack (attack detectability). Next, for a certain class of detection schemes monitoring a process, the relationship between the PCS design parameters, the closed-loop stability of the attacked process, and the detectability of certain attacks is rigorously characterized. Based on the characterization, for attack detection, it may be preferred to operate the process under performance degrading ``attack-sensitive'' parameters. To manage a potential tradeoff between attack detection and closed-loop performance, an active detection method utilizing switching between two control modes is developed. Under the active detection method, extended process operation is under a first (nominal) mode, the control parameters (called nominal parameters) for which are selected to meet traditional control design criteria. Under the second (attack-sensitive) mode, the process is operated with attack-sensitive parameters. The process is operated under the attack-sensitive mode intermittently to probe the process for an ongoing attack. Control parameter switching on a process under steady-state operation may induce transient behavior, which may trigger false alarms in the class of detection schemes. For processes with an invertible output matrix, a switching condition is imposed to select control parameter switching instances such that false alarms in the system are minimized. To eliminate false alarms due to control switching on processes with a non-invertible output matrix, a reachable set-based detection scheme is developed. The reachable set-based cyberattack detection scheme guarantees a zero false alarm rate during transient attack-free process operation by tracking the evolution of the monitoring variable values with respect to their reachable sets of the attack-free process at each time step. Following this, a switching-enabled active detection method that utilizes the reachable set-based detection scheme to enable attack detection with a zero false alarm rate is presented. Furthermore, the control parameter switching instances between the nominal to attack-sensitive modes are randomized, thereby preserving the confidentiality of the detection method. Destabilization of a process for attack detection (as with operation under attack-sensitive mode) may not always be preferred. Two different alternate control modes that may be used to induce perturbations for active attack detection without destabilizing the attacked process are presented. To guarantee attack detection, the alternate control mode selected must induce ``attack-revealing'' perturbations in the process. Reachability analysis is used to present a set-based condition that if satisfied means that the control mode selected induces attack-revealing perturbations. Different models of false data injection attacks are considered. A screening algorithm that may be used to select an attack-revealing control mode for the active detection of attacks is presented. The application of all methods are applied to simulations of different illustrative processes to demonstrate their attack detection capabilities.

Data-intensive computing has led to the emergence of data centers with massive processor counts and main memory sizes. However, the demand for shared resources has surpassed their capacity, resulting in increased costs and limited access. Commodity hardware, although accessible, has limited computational resources. This poses a challenge when performing computationally intensive tasks with large amounts of data on systems with restricted memory. To address these issues, Approximate Computing offers a solution by allowing selective solution approximation, leading to improved resource efficiency. This dissertation focuses on the trade-off between output quality and computational resource usage in sorting and searching problems. It introduces the concept of Approximate Sorting, which aims to reduce resource usage while maintaining an accepted level of sorting quality. Quality metrics are defined to assess the ”sortedness” of approximately sorted arrays. The dissertation also proposes a general framework for incorporating approximate computing into sorting algorithms, presenting an algorithm for approximate sorting with guaranteed upper bounds. The algorithms operate under a constraint on the number of comparisons performed. The dissertation continues to explore searching algorithms, specifically binary search algorithms on approximately sorted arrays. It addresses cases where metrics are given for the input array and cases where metrics are not available. Efficient and optimal algorithms are developed for multidimensional range searches and catalog searches on approximately sorted input. The dissertation further proposes algorithms that analyze patterns in input order to optimize sorting. These algorithms identify underlying patterns and sequences, facilitating faster sorting approaches. Additionally, the dissertation discusses the growing popularity of approximate computing in the field of High-Performance Computing (HPC). It presents a novel approach to comparison-based sorting by incorporating parallel approximate computing. The dissertation also proposes algorithms for various queries on approximately sorted arrays, such as determining the rank or position of an element. The time complexity of these querying algorithms is proportional to the input metric. The dissertation concludes by emphasizing the wide range of applications for sorting and searching algorithms. In the context of packet classification in router buffers, approximate sorting offers advantages by reducing the time-consuming sorting step. By capping the number of comparisons, approximate sorting becomes a practical solution for efficiently handling the large volume of incoming packets. This dissertation contributes to the field of approximate computing by addressing resource limitations and cost issues in data-intensive computing. It provides insights into approximate sorting and searching algorithms, and their application in various domains, offering a valuable contribution to the advancement of efficient, scalable, and accessible data processing.

The internet hosts an immense reservoir of videos, witnessing a constant influx of thousands ofuploads to platforms like YouTube every second. These videos represent a valuable repository of multimodal information, providing an invaluable resource for understanding audio-visual-text relationships. Moreover, understanding the content in long videos (think 2 hours), is an open problem. This thesis investigates the intricate interplay between diverse modalities—audio, visual, and textual—in videos and harnesses their potential for comprehending semantic nuances within long videos. My research explores diverse strategies for combining information from these modalities, leading to significant advancements in video summarization and instructional video analysis. The first part introduces an approach to synthesizing long video textures from short clips by rearranging segments coherently, while also considering audio conditioning. The second part discusses a novel technique for generating concise visual summaries of lengthy videos guided by natural language cues. Additionally, we focus specifically on summarizing instructional videos, capitalizing on audio-visual alignments and task structures to produce informative summaries.To further enrich the comprehension of instructional videos, the thesis introduces a cutting-edgeapproach that facilitates the learning and verification of procedural steps within instructional content, empowering the model to grasp long and complex video sequences and ensure procedural accuracy. Lastly, the potential of large language models is explored for answering questions about images through code generation. Through comprehensive experiments, the research demonstrates the efficacy of the proposed methodologies, envisioning promising future prospects in the field of semantics in long videos by integrating audio, visual, and textual relationships.

The EcoCAR Mobility Challenge is a four-year competition that challenges 12 teams in America and Canada to re-engineer a 2019 Chevrolet Blazer into a hybrid electric vehicle with Connected and Automated Vehicle (CAV) technology. This paper discusses The Ohio State University’s development of a perception system capable of detecting vehicles in the host vehicle lane, and adjacent lanes. A sensor suite consisting of six radar sensors, and one vision sensor was chosen. Simulations were performed in MATLAB to determine best sensor placements and orientations. Front sensor CAN data was obtained from real sensor hardware, decoded, analyzed, and plotted on a team-developed visualization tool in MATLAB. Radar and camera data were fused together using filtering and association algorithms to generate longitudinal and lateral track information about all vehicles inside the front sensors’ field of view. Persistent tracks were identified, tracked, and assigned unique track IDs. Velocity and acceleration were calculated, and a lane assignment algorithm assigned lane numbers to these tracked vehicles. The perception system algorithms – sensor fusion, tracking, and lane assignment – were calibrated, tested, and validated in four testing environments. The performance of the algorithms was evaluated over a period of 12 months using track-specific metrics.

A lot of recent work in the domain of privacy policy regulation has been around the choice of opting-in versus opting-out, a few others focus on the readability and length of the privacy policy, and some others have focused efforts to analyze network traffic. However, very few of the recent works in this field have examined the content of privacy policies. This work analyzes the content of privacy policies of mobile apps, application suites, social networks and ISPs. For each of the privacy policies examined in this study, this report outlines the choices that are offered to users over collection, use and sharing of personal, behavioral and non-personal information. This study sheds light on both direct and indirect sources of user data collection, and finally compares the privacy practices of the companies considered in this study, both within each category and between different categories.

Can business play a role in building resilience among vulnerable communities? Can (marketing) employment contribute to resilience - and if so, how? These are increasingly important and relevant questions in our current world as the recent COVID-19 pandemic showed many households to be just one shock away from falling into poverty. Consequently, understanding how resilience, the ability to cope both psychologically and economically with adversity, can be built has become a top priority for many policymakers and practitioners. This paper explores a marketing solution to building resilience. In this paper, we seek to answer the following question: "Can sales employment help people build resilience and enable them to cope with adversity?" We offer a mechanism through which sales improves resilience: by facilitating opportunities to connect. We show that by enabling opportunities to connect with a wide set of individuals, sales employment can build resilience. We differentiate between types of connections and show that different connections can lead to different forms of resilience. We present experimental evidence from a randomized controlled trial involving saleswomen in rural India. With our partner organization, we recruited, trained, and randomly assigned 1,048 women to either treatment, placebo, or control condition. We exogenously assign individuals to an employment position that facilitates opportunities to connect and measure the impact on resilience. In doing so, not only do we vary opportunities to connections, but we also provide a plausible mechanism of why they have more connections than others: certain jobs - in our case sales employment. To the best of our knowledge, there is no empirical evidence showing the impact of (changing) connections on resilience when facing adversity. In doing so, we are the first to randomly assign participants to different employment forms and causally estimate the impact of employment on personal outcomes, such as resilience. Our results show that those engaged in sales employment are more resilient than those in placebo employment or no-employment conditions. In fact, we find that sales employment significantly improves one's ability to be economically resilient when facing adversity by increasing: i) the likelihood of being aware of external sources of help and ii) the ability to take-up external sources of help. Second, our results show that engagement in sales is associated with having a wider set of connections. This effect is largely driven by increased connections with local government representatives and local community leaders. We find a strong association between the width of connections and economic and psychological resilience. Last, we examine heterogeneous effects and find that those from lower caste benefit most from engaging in sales employment in terms of building resilience compared. To the best of our knowledge, no one has looked at the impact of employment on resilience - we show that marketing employment holds the power to improve resilience when facing adversity.

Kenya has had a history of promoting both family planning and efforts against gender-based violence. However, intimate partner violence (IPV) remains a significant factor in women’s lives. Although experience of IPV has been shown to create a significant burden on reproductive health outcomes, less work has focused on the impact of IPV on women’s contraceptive use particularly in Kenya. Using the data from the 2003, 2008-09, and 2014 Kenya Demographic and Health Surveys (Kenya National Bureau of Statistics (KNBS) & ICF Macro 2004, 2010, 2015) this dissertation focused on the association between IPV, fertility intentions, women’s autonomy, and contraceptive use. I integrate the theory of gender and power and the social-ecological model to examine the association between IPV and contraceptive use.In the first study I focus on the impact of IPV on women’s fertility intentions. IPV experience increases the likelihood women would want to limit their childbearing as compared to wanting to space births or have children soon. This indicates that women would be less likely to want to raise children in an environment they are unsure is safe or supportive, a concept I call the uncertain futures hypothesis. The second study examines whether exposure to IPV reduced contraceptive use. I find that IPV did not uniformly reduce contraceptive use. Instead, IPV exposure decreased the likelihood of traditional contraceptive use but increased the likelihood of modern contraceptive use. This lends further credence to the possibility that women are making planful contraceptive decisions in IPV situations. In the final study I examine whether healthcare decision making, representing a facet of women’s autonomy, mediates the relationship between intimate partner violence and recent modern contraceptive use. This study found that there no mediated effect of healthcare decision making. However, healthcare decision-making remained significant in all models underscoring the need to consider ability to make healthcare decisions as a possible barrier to contraceptive access for women in abusive partnerships. This dissertation improves the evidence that IPV may facilitate women’s greater use of modern contraceptives in African contexts. It also underscores the need to consider IPV experiences when advising women on family planning.

Abstract: Cancer cachexia (CC) is a multifactorial syndrome characterized by severe depletion of skeletal muscle, with or without fat loss. A majority of cancer patients with incurable cancer are affected with cachexia and it remains an unmet medical need. Gene expression changes at the transcriptional level have been studied for CC in both humans and rodents. However, post-transcriptional regulatory mechanisms likely play a role in CC and are not addressed in the literature on human subjects, and are the focus of this thesis. It is well known that the sex-specific expression of genes contributes to traits and phenotypes. The role of autosomal expression of genes and their regulation at a post-transcriptional level in skeletal muscle from cancer cachexia patients has not been explored to account for the sex-specific differences in cachexia prevalence or severity. I hypothesize that CC, a manifestation of the net effect of complex host-tumor interactions and underlying molecular mechanisms, involves dysregulation of miRNAs and splice variants in skeletal muscle contributing to the pathophysiology of CC in a sex-specific manner. The objectives of the thesis are (i) to profile miRNAs expressed in skeletal muscle biopsies and to identify Differentially Expressed (DE) miRNAs associated with human CC using Next Generation Sequencing (NGS), (ii) to profile Alternatively Spliced Genes (ASGs) from human skeletal muscle biopsies and to identify differentially expressed alternatively spliced genes (DASGs) associated with human CC using Human Transcriptome Array 2.0 (HTA 2.0) and (iii) to profile and identify sex-specific miRNAs and Differentially Expressed Isoform specific gene expression (DEI) for human CC. Study subjects were classified based on the international consensus definition and diagnostic criteria for CC. Forty-two cancer patients were classified into 22 cachectic cases and 20 non-cachectic cancer controls who had no weight loss over a period of six months compared to their pre-illness weight. For identifying sex-specific expression differences, 22 cases and 20 controls were further stratified by sex. Body composition analysis was carried out using computed tomography (CT) taken prior to surgery. Total RNA was isolated from rectus abdominis muscle and used for NGS and HTA 2.0 arrays. Representative signatures were validated using qRT-PCR for cross platform concordance. Partek Genomics Suite v6.6 and SPSS v16 were used for bioinformatics and statistical analyses, respectively. Ingenuity pathway analysis (IPA) was used to identify canonical pathways and network analysis for biological interpretation. 777 miRNAs were profiled and 82 miRNAs with read counts of ≥ 5 in 80% of samples were retained for analysis. Eight miRNAs were DE (up-regulated, fold change of ≥ 1.4 at p < 0.05). IPA identified pathways related to myogenesis and inflammation. qRT-PCR analysis of representative miRs showed a similar direction of effect (p < 0.05), as observed in NGS. 8960 ASGs were identified, of which 922 DASGs (772 up-regulated, 150 down-regulated) were observed with > 1.4 fold change and p < 0.05. Representative DASGs validated by semi-quantitative RT-PCR confirmed the primary findings from the HTA 2.0. Identified DASGs were associated with myogenesis, adipogenesis, protein ubiquitination and inflammation. Up to 10% of the DASGs exhibited cassette exon (exon included or skipped) as a predominant form of AS event. Other forms of AS events such as intron retention and alternative promoters were also observed. Sex-specific expression was observed both at the miRNA and at the DEI levels. 10 upregulated miRNAs in males and 3 upregulated miRNAs in females were identified with ≥ 1.4 FC and p < 0.05. 1324 and 372 DEIs were identified in males and females, respectively. Pathways associated with skeletal muscle atrophy were predominant in males, whereas adipogenesis pathways were more common in females. Although the same upstream regulators were identified, their downstream targets were different between sexes. In all, my study has shown for the first time that post-transcriptional regulatory molecules are indeed associated with the pathophysiology of CC and are expressed in a sex-specific manner. This may potentially spur interests to address finer molecular mechanisms as an approach to personalized therapeutics development based on sex. Independent replications of the study findings are warranted. Functional characterization of the identified molecules may offer insights into their biological significance in cancer cachexia.

Previous studies produced the C8 exposure assessment which included PFOA release assessment, integrated fate and transport modeling, and dose reconstruction to predict the serum PFOA concentration of each individual in the C8 Health Project population from 1951 to 2008. The serum concentration predictions were used in various C8 Science Panel epidemiological studies to evaluate whether there is a ‘probable link’ between PFOA exposure and health effects. One such study analyzed the association with preeclampsia among the participants and found a moderate association (Savitz et al., 2012a). For this study, uncertainties in spatiotemporal predictions of PFOA water/air concentrations and in individual-level variables used in the dose-reconstruction and pharmacokinetic models likely resulted in some exposure measurement error, potentially affecting the validity of the epidemiological findings. The main objective of this dissertation was to analyze the impacts of different sources of input parameter uncertainty/variability in the C8 exposure assessment and study the impacts on the serum PFOA concentration predictions and the epidemiological association with preeclampsia. Monte Carlo (MC) simulation techniques were used to conduct these uncertainty analyses in this exposure assessment-environmental epidemiology model system.I found that autocorrelated and shared uncertainty in the PFOA water concentration estimates produced a highly variable set of plausible serum PFOA concentrations for study participants; however, it had less impact on the AOR of preeclampsia occurrence. Together, inter-individual variability and epistemic uncertainty in independent individual-level exposure parameters including water ingestion rates, the serum PFOA half-life, and the volume of distribution for PFOA impacted the serum PFOA concentration predictions and their association with preeclampsia moderately, with a 25% bias towards the null in the AOR of preeclampisa occurrence. Geocoding based uncertainty in residential addresses and work history together moderately impacted the rank exposure among the participants and caused a 41% bias away from the null in the AOR of preeclampsia occurrence. Future studies with complex exposure scenarios and multiple sources of variability/uncertainty might benefit from our approach of separating out the different kinds of uncertainty to better understand their individual impacts on the validity of epidemiological associations

Wide band gap (WBG) devices are the latest trend in the power electronics industry. MOSFETs built using WBG materials offer numerous benefits to power electronic circuits. The benefit of using WBG MOSFETs over conventional Si IGBTs is the ability to operate at higher switching frequencies and lower dead times, which reduces the switching losses. This thesis will explore the application of WBG MOSFETs to motor drives, where higher switching frequencies reduce harmonic losses and torque ripple and allow higher control bandwidth, Two-level voltage source inverters (VSI) utilizing Silicon Carbide (SiC) MOSFETs are constructed to operate a 1 HP induction motor. Experimental results are presented which show that the short turn-ON and turn-OFF times as well as high switching frequencies lead to increased shaft voltage and ground currents. Mitigation techniques of open-end winding drive and clamp-on ferrite chokes are evaluated in this thesis. The shaft voltage and ground currents are found to be eliminated when an open-end winding drive combines with the clamp-on ferrites. Simulation results of loss analysis is presented for a conventional two-level VSI and dual two-level VSI to understand the thermal requirements of the Si and SiC based drives. The advantages and disadvantages of WBG based conventional two-level inverter and the various mitigation techniques are discussed.

India, the world’s most populous nation, also had the highest absolute number of Tuberculosis (TB) cases in 2011. The Revised National TB Control Program (RNTCP) has implemented the Directly Observed Treated Short Course (DOTS) strategy throughout the country but with suboptimal population coverage with the remainder cared for by the private sector. Treatment success rates above 85% have consistently been reported by the national TB program, but independent evaluation has been limited. The aim of this thesis was to study DOTS program performance in South India and to estimate infection rates and prevalence while accounting for exposure risk. The utility of the QuantiFERON Gold In-Tube test (QFT) to detect TB infection in this setting was also examined. A cross sectional review of treatment records of TB patients along with prospective follow-up of a cohort of TB patients were used to evaluate DOTS performance. In addition, the prevalence of infection and disease at baseline as well as after 6 months of follow-up were examined in household contacts of the cases and neighbourhood community controls, with use of both tuberculin skin tests (TST) and QFT. High treatment success rate (93%) among pulmonary TB patients was recorded in the chart review but with a high rate of missing outcome data, while the treatment success rate was low (69%) in the prospective study with frequent deaths (12%) and defaults (19%). Elderly TB patients experienced poor treatment outcomes in this setting. Prevalence and incidence of TB infection was high in both contacts and controls with some risk associations identified including smoking (active and passive) and increasing proximity to index TB cases. Test coverage with QFTs was better than the TSTs at follow-up but these logistic benefits were offset by the high test cost, laboratory demands and the lack of clear conversion definitions. This thesis highlights shortcomings in RNTCP program performance in South India, including inadequate record keeping practices and a low treatment success rate. These findings argue for additional support for high risk individuals and better record management, which should be a focus of future research.

Spinal Muscular Atrophy (SMA) is an autosomal recessive neurodegenerative disease characterized by loss of lower motor neurons and muscle atrophy. It is caused due to reduced levels of the ubiquitously expressed Survival Motor Neuron (SMN) protein. Humans have two SMN genes – SMN1 which produces full length SMN and SMN2 which mainly produces a truncated unstable protein and a small amount of full length SMN (~10%). SMN is a multi-functional protein with a known role in snRNP assembly. To date, it is still not known why the lower motor neurons are specifically affected in SMA although SMN is expressed in all tissues.Several missense mutations of SMN1 have been studied to understand the mechanism of SMA. One such synthetic mutation which affects the first glutamine (Q) of the QNQKE motif present in exon 7 is Q282A. A previous study has shown that this mutant construct cannot rescue the axonal defects in a zebrafish model of SMA. Since in vitro assays have suggested that SMN-Q282A can function in snRNP assembly, the authors proposed that restoring snRNP assembly function of SMN was not sufficient to rescue SMA. To analyze this mutant further, we generated transgenic mice carrying SMN-Q282A and crossed them to generate SMA mice with the following genotype: SMN-Q282A;SMN2;mSmn-/-. Currently, we have 25 such rescue mice and they show no physical or behavioral abnormality when compared to normal mice. This indicates that SMN-Q282A acts to complement the full length SMN produced by SMN2. However, as in previous studies with SMN missense mutations, the SMN-Q282A could not rescue a mSmn-/- mice. This is to be expected since no single mutant form of SMN has been able to rescue a mSmn-/- mouse by itself. These results suggest that snRNP assembly plays an important role in SMA and that the axonal defects in fish do not correlate with the critical function of SMN in SMA. We are currently performing snRNP assembly assays on brain and spinal cord lysates of these rescue mice. This will confirm our hypothesis that restoring snRNP assembly is sufficient to rescue SMA mice. Similar experiments were done with SMN-I116F, a mutation seen in SMA Type I patients. Initial results show that this mutant construct can provide only minimal complementation to SMN2. SMA mice of genotype SMN-I116F;SMN2+/+;mSmn-/- could survive only for 16 days, similar to the SMND7 SMA mice. The effect of another mutation SMN-D44V, seen in a Type III patient, was studied using scAAV9 vector. The viral vector carrying the SMN-D44V mutant was injected via the facial vein into SMA neonate pups at day 1. Partial recovery was seen with survival of the mice extended to 35±2 days. Transgenic mice having the SMN-D44V mutant have been generated and we expect them to fully complement SMN2, being a milder allele. Further experiments are being carried out to characterize these two transgenes, which will give us more input as to how different mutations affect the functioning of the SMN complex and its composition.

Implantable systems are used in various contexts for interfacing with the body and for providing real-time monitoring and control capability. In particular, implantable neural interfaces can be used to radically improve our understanding of the nervous system and to provide precise treatments for a variety of neurological problems. However, these systems require significant computing power to perform real-time in-situ analysis of neural signals to recognize behaviorally meaningful patterns which are used to trigger appropriate corrective actions. Due to the tight area and power constraints of neural implants, it is important to develop novel algorithm-architecture-circuit co-design approaches for efficient implementation of neural signal analysis. First, we develop an algorithmic framework which is suitable for ultralow-power hardware implementation while simultaneously satisfying emerging design requirements like reliability and security. The algorithm is based on building a dynamic hierarchical multi-level vocabulary of neural patterns in the wavelet domain. The vocabulary-based analysis allows recognition of neural patterns at multiple levels (spike, burst, and pattern of bursts across multiple channels) and transmission of recorded data with large compression, thus, saving power and communication bandwidth of the integrated telemetry device. Hardware implementation of the proposed algorithm aims at reducing system power through choice of appropriate architecture and circuit-level design techniques. We show that a super-threshold design operating at a much higher frequency can achieve comparable energy dissipation as a sub-threshold low-frequency design through application of extensive power gating. It also provides significantly higher robustness of operation and yield under large process variations. We propose an architecture-level preferential design approach for further energy reduction at the cost of graceful degradation in output signal quality under voltage scaling and parameter variations. Considering the emerging need of secure computing in implantable systems, we analyze the various security threats in the proposed system. We exploit the vocabulary-based encoding of neural signals to realize an ultra-lightweight data obfuscation solution. Furthermore, we consider an emerging security threat namely, hardware Trojan attack, where an adversary introduces malicious modifications of a circuit during design or fabrication. We analyze the effectiveness of different Trojan attacks in implantable systems and develop side-channel analysis based Trojan detection approaches.

Hardware Security is a key issue with the emergence and proliferation of embedded systems. Many embedded systems like smart cards, RF-id tags, mobile phones and PDAs run cryptographic algorithms running on dedicated cryptographic ICs to perform operations like secure identification, authentication and communication. Even though cryptographic algorithms are mathematically strong and secure, the underlying hardware (CMOS) implementations may reveal the secret key of the device by exhibiting data or operation dependent power consumption. Attack methodologies that use information like power consumption, timing information and electromagnetic radiation of the Cryptographic IC to find the secret key of the device are called Side Channel Attacks.Differential Power Analysis (DPA) is a category of Side Channel Attack that uses differences in power consumed for different input vectors along with statistical analysis to reveal the secret key in the device. DPA attacks are possible due to the data dependent power consumption of the CMOS circuits. Several algorithmic, circuit and logic level countermeasures have been proposed to resist DPA attacks. Unfortunately many of these techniques are either ineffective or end up having a huge overhead in other design metrics like area, power and speed. The objective of DPA resistance is to remove data dependence from power consumed. Recently, many circuit design techniques based on Dynamic and Differential Logic, which allows for matching power profiles regardless of input, have been proposed.In this thesis, we present a universal cell called SDMLp (Secure Differential Multiplexer Logic), based on differential pass transistor logic, to mitigate DPA attacks. A single two input SDMLp cell can be configured to perform all possible 16 two input logical operations. Employing SDMLp cells in circuit design increases the DPA resistance of the circuit by exhibiting 10 to 100 fold reduction in instantaneous power variance compared to existing DDL based secure logic styles and nearly 1000 fold compared to SCMOS designs. Apart from improvement in instantaneous current variation, we also show nearly 60% to 70% reduction in power consumption and nearly 45% to 50% reduction in area compared to existing DDL based secure logic styles.

Previous studies have shown that the application of High Frequency Alternating Currentto nerve fibers results in a reversible quick acting conduction block. The commercialization pathway for the application of HFAC technology to treat chronic postsurgical pain in patients has been explored. The technical portion of the thesis provides a general overview of the nervous system, mechanism of pain, neuromodulation and its therapeutic application as well as a detailed explanation of the HFAC technology. Thebusiness portion of the thesis explores the commercial feasibility of the technology against market, competitive landscape, FDA regulatory pathway, reimbursement, intellectual property, financing required and potential exit strategies.

One of the potential sources of vibration during rolling of a non pneumatic tire is the buckling phenomenon and snapping back of the spokes in tension when they enter and exit the contact zone. Another source of noise was hypothesized due to a flower pedal ring vibration effect due to discrete spoke interaction with the ring and contact with the ground during rolling as the spokes cycle between tension and compression. Transmission of vibration between the ground force, ring and spokes to the hub was also considered to be a significant contributor to vibration and noise characteristics of the Tweel. Previous studies have studied spoke vibration, ground vibration and related geometrical factors on a two-dimensional (2D) Tweel model. In the present work, a three-dimensional finite element model of a non-pneumatic tire (Tweel) was considered which uses a hyperelastic Marlow material model for both ring and spokes based on uni-axial test data for Polyurethane (PU). Changes in material properties on static load-deflection curves and vibrations of spoke and ground force reaction during high-speed rolling are studied. In addition, energy loss upon impact with an obstacle is also studied. For static load deflection studies, a new analysis procedure is developed which allows for a cooling step to proceed prior to loading, and yet maintains continuous contact with the ground. For the dynamic rolling studies, a direct analysis procedure is developed, where the Tweel is accelerated from rest. This procedure avoids potential numerical difficulties when defining nonzero initial speeds as used in previous studies. In order to study the effect of changes in shear modulus for the ring and spokes while keeping the ratio of volumetric bulk modulus to shear modulus unchanged, the value of shear modulus is varied from Mooney-Rivlin and Neo-Hookean models obtained from a least-squares fit of the uni-axial stress-strain data. A total of 6 different material models are examined together with the original Marlow model. The 6 material models are divided into 2 sets and each set has 3 levels (unchanged and plus/minus 25% change in shear modulus). Upon evaluation of the uniaxial data, the results show that on increasing the shear modulus, the tangent slope of the normal stress-strain curve increases; whereas with decreasing shear modulus, the slope decreases. For tensile stresses and strains, the Mooney-Rivlin best matches the original Marlow material model, compared to the simpler Neo-Hookean model. However, for large compressive stresses, the Mooney-Rivlin diverges significantly from the Marlow curve. The simple Neo-Hookean model is able to fit the Marlow curve better for compression, but is less accurate in tension. As a result of decreasing shear modulus, the vertical displacement in the static load-deflection curves increases upon loading. The Neo-Hookean model resulted in decrease in stiffness when compared to the Mooney-Rivlin and original Marlow model. The effects of material changes on spoke vibration as measured by changes in perpendicular distance and vibration in ground interaction measured by FFT frequency response of vertical reaction force during rolling are also reported. Results show a trend the vibration decreased when the stiffness of the Mooney Rivlin and the Neo Hookean models was increased from +25% to -25%. Conversely, the vibration increased when the stiffness decreased between the extreme limits. However, in several of the material models for the ring and spokes, the unchanged stiffness gave the lowest vibration amplitude, suggesting that a optimal value is somewhere between the plus/minus 25% stiffness limits. To study energy loss the 3D finite element model of the Tweel is rolled over an obstacle whose height is 7.5% of the radius of the Tweel. Energy loss is measured by the reduction in axial hub velocities and kinetic energies (KE) relative to an analytical rigid wheel with the same mass, moment of inertia and initial velocity. Results show that the reference Tweel with Marlow material properties, after traversing the obstacle, resulted in an average reduction in axial velocity and total kinetic energy of only 1.3% and 2.3%, respectively. Results show that for Mooney Rivlin, a decrease in shear modulus caused a decrease in energy loss. Conversely, for Neo Hookean, a decrease in shear modulus resulted in an increase in energy loss and an increase in shear modulus resulted in a decrease in energy loss.

The number of skin cancer cases reported in the United States is increasing every year and nearly equals the total cancer cases detected from every other part of the body. Current detection strategies of skin cancers include a visual examination followed by a tissue biopsy. This procedure is subjective, invasive and time-consuming. Therefore, considering the number of cancer cases reported and the number biopsies performed, there is a critical need for a non-invasive diagnostic aid to help clinicians reduce the significantly large numbers of unnecessary biopsies. This dissertation presents a quantitative method based on optical spectroscopy for performing a non-invasive ‘optical biopsy’ of melanoma and non-melanoma skin cancers. We have developed the hardware, software and optical algorithms necessary to implement such a device. First, we present a novel lookup table-based model for determining the optical properties of tissue that is valid for fiber-based probe geometries with close source-detector separations and in highly absorbing tissue. These optical properties are quantitative parameters that can be correlated with the physiology of tissue. Second, we present experimental validation of the effects of microvasculature pigment packaging on diffuse reflectance spectra. We have conducted experiments using microfluidic devices over a physiologically relevant range of optical properties and blood vessel sizes. Third, we present the development of a probe-based portable and clinically compatible instrument capable of in vivo spectral measurements. The instrument combines two modalities – diffuse reflectance and intrinsic fluorescence spectroscopy – to provide complementary information regarding tissue morphology, function and biochemical composition. Finally, we present the results of a pilot clinical study using our portable instrument to determine the accuracy of spectral diagnosis of non-melanoma skin cancers. Our results show that the mean optical properties and fluorophore contributions of normal skin and non-melanoma skin cancers are significantly different from each other and can potentially be used as biomarkers for non-invasive diagnosis of skin cancer.

CD4+T-cells have a central role in induction and homeostasis of the immune response, and are also the major target cells for HIV. HIV has devised mechanisms to subvert the immune system and further its cause of survival and dissemination. vpr is one of the accessory genes, which is essential for the virus survival in vivo. Being a soluble protein with an ability to transduce across cell membranes, Vpr can potentially affect bystander cells. We hypothesize that HIV-1 Vpr alters the functions of both infected and bystander T lymphocytes, utilizing direct and indirect mechanisms, and these Vpr-mediated effects contribute in part to the immune dysregulation, and aid in viral dissemination. The Specific Aims of this proposal are to: (1) Assess the immune modulatory effects of Vpr in infected and bystander T-lymphocytes; (2) Understand the role of Vpr in T lymphocytes, natural killer (NK) cells and dendritic cells (DC) interactions; and (3) Analyze the structure-function role of Vpr in immunopathogenesis. We utilized HIV-1wt and HIV-1ΔVpr viruses and compared the difference in the effects of these viruses under controlled simulated invitro conditions. The differences observed in the effect of these two viruses can be attributed to Vpr provided that the infections in both the experimental sets are similar. In some studies, to clearly identify infected cells, we employed EGFP reporter viruses. The effects in infected cells and bystander cells were evaluated. Results indicate that HIV-1 Vpr has a role in dysregulation of immune cells during HIV infection. Vpr differentially regulates the surface expression of T cell costimulatory molecules, CD28 and CTLA-4, and inhibits IFN-γ production in infected T cells. Vpr also inhibits NK cell function by augmenting TGF-β production and inducing altered expression of NK receptor ligands. Further, oligomerization of Vpr has a role in gag incorporation of Vpr and in viral pathogenesis. The findings presented in this study are significant for public health because mechanistic understanding of the pathogenesis will aid in development of novel anti-viral therapeutics.

An experimental process for obtaining fibers and spheres using electrospinning of a polymer (Polyethylene Oxide) solution in water were studied. The objectives of the study were to complete measurements of size of the generated fibers and spheres to develop phenomenological analyses to help interpret and correlate the measurements of the instability growth on the electrospun micro liquid (polymeric solution) jets. Digital holographic microscopy was used to observe the trajectory of the electrospun jet and the growth of instability on the electrospun micro polymeric jet. The morphology of the fibers deposited on the collector electrode was observed using SEM imaging. The conditions for formation of beaded fibers were studied. The sphere formation mechanism was observed for different parameters including polymer concentration, applied voltage, and needle to electrode distance.

In this work, a new MEMS technology for hot embossing using polydimethylsiloxane (PDMS) tools is developed and characterized. The developed technology will permit low cost, simple, and rapid fabrication of disposable microfluidic biochips and plastic microstructures for numerous BioMEMS applications. For traditional planar systems, a negative photoepoxy (SU-8) or thick positive photoresist (AZ4620) on silicon were used to form micromolds, upon which PDMS was cast. Fabrication time of these PDMS tools was considerably less than that of conventional ones made from nickel or silicon, and remained the same regardless of the feature aspect ratios. The developed PDMS tools were used to rapidly prototype microchannels and microfluidic biochips in polymethylmethacrylate (PMMA) of aspect ratios up to two, with depths ranging from 5 to 250 µm, and features as small as 40 µm in width. The use of a soft PDMS tool material increased cycle time and embossing force, and limited the tool lifetime to approximately 20 cycles. The technique was also extended to fabrication of orthogonal 3-D microfluidic systems using multiple lithography steps. In addition to microchannels of the traditional planar systems, these 3-D systems contained features on the microchannel bottom surface, providing obstructions in the flow field. The orthogonal 3-D PDMS tools were fabricated by casting in a micromold formed on silicon substrates using a combination of SU-8 and AZ4620 photoresists. This approach proved to be a much simpler and faster process that the conventional multi-step nickel electrodeposition method. Overall, the developed technique enables simple and low cost fabrication of planar and orthogonal 3-D microfluidic systems, and is particularly well suited for rapid prototyping applications, where fast processing and low cost are important and the number of samples is relatively low. Combined with a high-resolution mask printing on transparencies, a complete microfluidic device could be fabricated in less than 24 hours.

Simulation has been an integral part of the VLSI development cycle that has been in practice today. VHDL-AMS is a language that is well standardized for time domain simulations and verifications. But it has become increasingly important to analyze and simulate models in frequency in an effort to be able to perform microwave and RF simulations. Hardware description languages and their simulators are needed to model and simulate mixed time/frequency systems. The requirement of having a simulator that performs frequency domain simulation along with time domain simulation has always been a major need and is necessitated by the enormous number of frequency variable circuits that have come to the market today. Although VHDL-AMS language provides the ability to model systems over a wide range of physical domains, it is primarily a time-domain language. As a result, frequency-dependent features are poorly supported. This thesis presents research in mixed time/frequency domain simulation where models, written in an extended VHDL-AMS (for frequency), can be specified and simulated in time, frequency or both. Furthermore the ability to convert models in time to frequency and simulate them in frequency, which is detailed in this thesis, increases the simulator's scope and functionality. Joint time-frequency analyses, the ability to analyze and simulate a model in both time and frequency, is also been dealt with in this thesis in an effort to present a complete general purpose simulator.

Since the beginning of the environmental movement, individuals and organizations have worked to enact changes that they hope will improve the state of the environment. A large portion of this movement is based on increasing the population’s environmental awareness and consciousness. Much of the conservation community believes that by increasing awareness and concern for the environment, they will indirectly increase conservation efforts. The logic behind this belief is that consumer’s knowledge, attitudes and concern for the environment are determinants of behavior and will manifest itself in their lifestyles and green purchasing decisions. The purpose of my study is to ascertain whether variables, specific to environmental consciousness are more suitable for characterizing consumers purchasing decisions. Specifically, measures of environmental knowledge, attitudes and behavior are linked to two conceptualizations of the purchasing domain, namely green purchasing decisions in general and the specific purchasing habits of five green categories (recycled paper products, not tested on animals, environmentally friendly detergents, organically grown fruits & vegetables and ozone- friendly aerosols).

There are many factors governing the performance of TCP-based applications on satellite channels. The end-to-end performance of TCP is known to be degraded by the delay, noise and asymmetry inherent in geosynchronous systems. These results have been based on experiments that evaluate the performance of TCP in single flow tests. Single flow tests are useful for deriving information on the theoretical behavior of TCP and allow for easy diagnosis of problems. However, they do not represent realistic situations and therefore cannot be used to authoritatively comment on performance issues. Our experiments tested TCP's performance in a more dynamic environment with competing traffic flows from hundreds of TCP connections running simultaneously across the satellite channel. Another aspect investigated was TCP reaction to bit errors on satellite channels. TCP interprets delays and losses as a sign of network congestion. This causes it to cut down its transmission rate leading to poor performance. We allowed the bit error rate on our satellite channel to vary widely and tested the performance of TCP as function of these bit error rates. Our studies show that the average performance of TCP on satellite channels in these realistic conditions is good and establishes the robustness of TCP on satellite channels. Many of our observations contradict commonly held beliefs about the performance of TCP on satellite links.

The purpose of the research has been to develop a mathematical-computer model for sound propagation in human thorax. One of the major advantage of this model is the incorporation of the visible human male data set to have an accurate representation of human thorax. The visible human male data set consists of CT images of a human body at a resolution of 1mm by 1mm by 1mm. Various types of simulations were performed at 4mm by 4mm by 4mm CT image resolution and acoustic pressures were sampled at four different points with in the thorax. The results were analyzed in time and frequency domains. The results from the analysis suggest that frequencies of the order of 100Hz are most effectively transmitted through the thorax and there seems to be a resonance effect at 1500Hz. Further analysis indicates that the resonance frequency could be due to the spatial confinement of the sound waves within the thorax. The model also exhibits a spatial inhomogeneity of sound propagation. The results confirm that the size of the thorax can play a significant role in the type of sound generated at the chest wall.

Emerging high performance, high speed and scalable networks allow machines to access remote memory quicker than accessing disks. This paves way for coordinating the file caches of workstations in a distributed local area network, known as Cooperative Caching. This study introduces a novel cache architecture for cooperative caching called “Reliable, Efficient and Distributed Cooperative Cache”. We present three algorithms – Server Logging, Client Logging and Peer Logging – that combine duplication with logging in order to improve the write performance of cooperative caching while ensuring the reliability of the data. We evaluate the performance of the logging algorithms using trace driven simulations. Several synthetic traces and two real world traces were used to measure the performance improvement obtained by the algorithms. The simulations indicate that the speedup obtained is as high as 10 compared to the baseline system. Based on the simulations, we conclude that the logging algorithms make complete use of the cooperative caching environment and provide better overall improvement by reducing the response time of write requests, improving individual client performance, minimizing the load on the server and ensuring reliability and availability of data.

The purpose of this project has been to implement a differentiated services (diffserv) router in software in a Linux environment. The objective is to enable the construction of a test-bed of such diffserv routers. One of the design goals is to make the software as modular as possible. This permits the various modules of the diffserv router to be configured easily. In addition, a very user-friendly script file interface was designed to configure the diffserv modules. Consideration was also given to the fact that in future, this code is expected to be integrated with a policy server. In this project, the emphasis was on functionality and not on performance. Hence, not much effort was taken towards optimizing the performance of the diffserv router.

Sheet metal stamping is the process of forming various panel shapes such as automotive doors, hoods etc. It includes a forming operation that sets the panel shape and dimension. The forming process is complex and requires successful interactions between its several input variables (blank, dies, drawbeads etc.) for yielding good quality panels. However, in current practice, there is insufficient monitoring for continually produced panels. Panel dimensional control is rarely done, resulting in quality problems in the final car assembly. Moreover, there is a lack of scientific process understanding, making the trouble-shooting of panel defects (splits, wrinkles etc.) difficult. A trial-and-error approach of the die makers is relied upon, which is time consuming and expensive. The primary contribution of this research is the development of a process monitoring methodology using the available output process signatures, namely, panel strain (epsilon) and punch force (f). Towards this objective, an adaptive strategy for efficient strain measurement has been developed, and in addition, a deeper understanding of the impact of drawbead faults has been made. The process monitoring methodology is a tool applicable in production environments for identification of the root cause that results in a defective panel. Pertinent information (or features) such as input work, peak strains etc. are extracted from the signatures to characterize the process. The methodology involves the development of a linear mapping between the input variables and the selected output features through systematic training. The trained mapping is later applied to the features from a defective panel for root cause identification. The methodology is demonstrated and validated for a cylindrical cup deep-drawing, hemispherical punch stretching and box drawing processes. The adaptive strategy for strain measurement involves etching uniform small square grids over the panel, and measuring strains adaptively based on the deformations of larger squares comprising many smaller grids. The adaptive scheme uses FEA strains as input and accounts for experimental noise. The superiority of the methodology is demonstrated in that the strain (both low and high) features are measured accurately, and in addition, measurement time is saved significantly. For a better process understanding in the presence of drawbead faults, a generalized 2D FEA modeling procedure is adopted to study the restraining forces in both straight and curved drawbeads. The results indicate that the restraining force for a curved drawbead is close to the sum of the contributions from an identical straight drawbead, and the metal flow-in deformation of the curved flange. In addition, the sensitivity of the restraining force for the curved drawbeads, is similar to that of the straight drawbeads.

Auditors are encountering more and more computerized accounting applications as the pervasiveness of computing technology increases in business. Auditors, therefore, need to adapt their audit approaches in the face of the changes caused by the new technology. The AICPA has addressed the issue by requiring auditors to consider the nature of the data processing system ln their client environments when planning the audits. Specialists, if necessary, are recommended to be brought in as part of the audit team in audits involving computerized accounting applications. The implicit assumption behind this is that the specialists would make “better” judgments in auditing computerized systems than non-specialists. A need was seen to compare the judgments of specialists and non-specialists in evaluating controls in a simple computerized environment. The results indicate that while both specialists and non-specialists have a high degree of consensus, a significant difference existed between the two groups of auditors. Both groups of auditors exhibited high reliability and self-insights. Experienced non-specialists had lower consensus than specialists while inexperienced non-specialists had lower reliability than specialists. Firm affiliation effects were noted for the non-specialists in their consensus scores. Unlike previous studies, segregation of duties cue did not account for a majority of the variance In judgments. This cue was considered important only by the experienced non-specialists. A need was seen for further research into how the difference ln consensus affects subsequent audit program planning.