Indian Institute of Science Bangalore
etd@IISc Electronic Theses and Dissertations at Indian Institute of ScienceNot a member yet
6204 research outputs found
Sort by
Design, Development and Testing of Cervical Dilatation Measurement System
No full textMonitoring labour progression is of utmost importance, to avoid complications during childbirth. Frequent monitoring allows prevention, identification, and management of complications during childbirth. Improper monitoring can result in serious complications and can negatively impact the health of both the mother and the baby. The problems get exaggerated in remote villages due to lack of medical infrastructure and timely transportation. The parameters that are commonly monitored during labour are frequency, strength and duration of uterine contractions, cervical effacement and dilatation, and foetal head station. Cervical dilatation is one of the most important parameters used to gauge the progress of labour. Monitoring cervical dilatation alone, allows healthcare professionals to take decisions regarding subsequent intervention.
Digital trans-vaginal examination, when conducted by an experienced practitioner, is considered the gold standard for obstetric practice. However, the measurement is inaccurate due to inter and intra-observer variability. The variability results from the fact that the cervix is soft and can get easily stretched during measurement. Patients, especially the ones with ruptured membranes, are put at heightened risk of infections when subjected to repeated vaginal examinations and they also experience discomfort.
In an attempt of improving the accuracy of measurement, different methods have been explored in literature. Various mechanical measurement tools that were devised were found to be heavy and caused distortion of the cervix. Ultrasonic monitoring methods require the sensors to be screwed into the tissue at the exact locations. This may lead to local trauma and any error in sensor placement could lead to increased measurement errors due to which this may not be the preferred method of measurement for cases of normal labour. While high-resolution, real-time ultrasound imaging systems are available, they are prohibitively expensive. None of the methods mentioned above have managed to reach a hospital setting.
The aim of this work is to develop a cervical dilatation measurement system which is accurate, cost-effective, safe, and easy to use, and causes least discomfort to the patient. It is desired that the sensing mechanism used for measurement be non-contact in order to reduce patient trauma. Hence, imaging using cameras is explored in this work. Through extensive literature study of the complex and dynamic anatomy of the female reproductive system, it is established that the vaginal canal may occlude any non-contact sensor placed at the vaginal introitus. In order to support measurement by any such sensor, a mechanism is required to retract the vaginal walls. Various existing vaginal retractors, also called specula, are evaluated and a novel retractor is designed to suit cervimetric applications, such that it provides maximal cervical visibility while causing minimal patient discomfort. The materials used in the manufacturing of the retractor and the manufacturing process is also discussed. This retractor is evaluated to be safe for use within the vaginal environment.
Images of the dilating cervix are studied. Various image processing techniques such as manual, colour-based, and edge-based detection methods are established and evaluated for the detection of the cervical os. A consolidated image processing tool is also developed and guidelines for choice of image processing parameters are established.
The specifications of the camera system required for use in cervimetry are determined. Two types of camera systems: monocular and stereo systems are developed in this work. The bio-compatible packaging of these systems is discussed. These systems are tested on cervical models and evaluated for use in cervimetry. It is found that endoscopic imaging in cervimetry performs better than most of the state-of-the-art cervimetric methods found in literature. Overall, a cervimetric product which utilizes an imaging device and a supporting retractor, along with associated image processing has been developed in this work with the intention of its long-term and sustained utilization in a clinical setting
Correlation Between Microstructure and Corrosion Properties of Electrodeposited Nickel and Zinc Based Coatings
No full textConventional metallic coatings such as Ni and Zn are extensively used in corrosion protection applications. In most studies on the corrosion behavior of metallic coatings, the emphasis is on correlating morphology and corrosion behavior. The focus on relating the microstructure (phase fraction and spatial location of phases), micro-texture, grain boundary constitution, and strain in the coatings with its corrosion performance remains relatively less explored. Corrosion initiates at the surface and propagates through the coating to reach the substrate-coating interface; therefore, the coating's microstructural attributes play a significant role in deciding the degradation kinetics.
In this work, such correlation between microstructure and corrosion properties has been investigated for the following systems: Ni and Zn coatings electrodeposited using surfactants of different polarity, Ni and Zn coatings electrodeposited with carbon-nanotubes (CNTs) as a foreign additive, Ni-Co alloy coatings with different concentrations of Co, and Zn-Ni alloy coatings with different concentrations of Ni. In all the cases, a direct correlation between the coating micro-texture, grain boundary constitution, strain, and corrosion behavior was noticed. This is the main conclusion of this research work. The key observations were: (a) in the case of Ni coatings deposited using surfactants of different polarities (cationic (CTAB) surfactant, anionic (SLS) surfactant, and non-ionic (Triton X-100)) surfactant, highest corrosion resistance was obtained in case of anionic (SLS) surfactants containing coating because of lesser density of geometrically necessary dislocations (GNDs) and lower coating strain, (b) in the case of Zn coatings electrodeposited using different surfactants (cationic (CTAB) surfactant, anionic (SLS) surfactant, and non-ionic (Triton X-100) surfactant), highest corrosion resistance was observed for coating with cationic (CTAB) surfactant due to a low fraction of high angle grain boundaries (HAGBs), a high number of low energy special boundaries, and relatively lesser defective morphology, (c) in the case of Ni-CNT composite coatings produced form electrolyte bath containing different concentrations of dispersed CNTs, highest corrosion resistance was observed for Ni coating produced from an electrolyte bath containing 50 mg/L of CNTs, this was due to high coincidence site lattice (CSL) fraction (greater than 40%) when compared to the other Ni-CNT composite coatings (d) in the case of Zn-CNT composite coatings produced form electrolyte bath containing different concentrations of dispersed CNTs, highest corrosion resistance was observed for Zn-CNT coating produced form the electrolyte bath containing CNT in concentration of 8 mg/L. This was due to the presence of low energy basal plane texture, low fraction of HAGBs and a high number of low energy special boundaries, (e) in the case of electrodeposited Ni-Co coatings with different concentrations of Ni, lowest corrosion rate was observed in the case of Ni-20 wt.% Co due to uniform distribution of Ʃ3 CSLs along the coating cross-section, (f) in the case of Zn-Ni alloy coatings containing different minor concentrations of Ni (1-4 wt%), high corrosion resistance was observed in case of Zn-1.34 wt.% of Ni due to the presence of γ-phase at the grain boundary regions and less micro-strain in coatings
Anthropogenic Influence on River Water Quality
No full textAnthropogenic factors such as climate change, land use land cover change and industrial and population growth can influence river water quality. Climate change affects water quality due to changes in stream temperature and streamflow due to increased air temperature and varied precipitation patterns associated with warming. Land use land cover influences water quality mainly from the agricultural runoff, which carries the pollutants from fertilizers and pesticides and reaches the nearby water body. Population growth can increase the water demand and sewage generated hence aggravating pollution. Industrial growth has the potential to affect water quality through increased effluent loads. The work presented in this thesis contributes to quantifying such anthropogenic influences on river water quality using a coupled hydrological-water quality simulation model. The study area considered is a 238km stretch of Ganga river in India from Ankinghat to Shahzadpur, passing through Kanpur, which is identified as the most polluted stretch of Ganga river by the Central Pollution Control Board of India.
Sensitivity studies with forcings such as climate change and land use are extremely important for any management decision on water quality. In the initial part of the thesis, the sensitivity of nine water quality parameters to climate change and land use change is assessed using idealized scenarios and a standalone water quality simulation model, QUAL2K. The key input model parameters contributing to model uncertainty and key locations are identified using first order reliability analysis. The water quality parameters considered are DO, BOD, ammonia, nitrate, total nitrogen, organic-, inorganic-, and total phosphorous and faecal coliform. The non-point source pollution is quantified using the export coefficient method, in which pollutants from all land use classes are considered. Eight climate change and six land use land cover scenarios are framed based on historical data analysis to assess their sensitivity to water quality parameters. DO is the most sensitive indicator to the climate change scenarios considered, while nutrients and faecal coliform are more sensitive to the land use scenarios. In general, the water quality parameters are found to improve with a rise in air temperature and deteriorate with a reduction in streamflow. An increase in the agricultural land area leads to higher nutrient concentration, while an increase in the built-up area causes an increase in faecal coliform concentration. An increase in forest land shows better water quality in terms of all water quality parameters. The key input variables contributing to the uncertainty of water quality simulation are the head water discharge, point and non-point pollution loadings, water temperature, and corresponding reaction rates. The key locations identified using first order reliability analysis are Kanpur downstream and Jajmau downstream.
Risk assessment studies on water quality for future scenarios are limited in the literature. In the next part of the thesis, the effect of climate change on water quality, the risk of eutrophication and fish kill for the mid-and end of the 21st century for this river stretch are assessed. The risk of eutrophication and fish kill are quantified using simulated concentrations of nutrients and DO, respectively. Downscaled climate change projections for two climate change scenarios (RCP4.5 and RCP8.5) are used to drive a hydrological model coupled with a water quality simulation model. The simulations indicate a potential deterioration of water quality in this stretch in the mid-21st century, with a potential increase in pollutant concentration by more than 50% due to climate change alone. The risk of reduced dissolved oxygen and increased organic and nutrient pollution, and the risk of eutrophication and fish kills increase with warming due to the rise in the frequency of low-flow events and a reduction in streamflow during low-flow events. However, the risk of nitrate and microbial pollution is reduced due to increased denitrification and pathogen decay rates with warming. The risk of eutrophication and fish kill is found to increase by 43.5% and 15% due to climate change alone by the mid-21st century. The risk of eutrophication is found to increase by 6% due to land use change which can be attributed to an increase in nutrient loading with land use change.
In the final part of the thesis, the individual effects of climate change, land use land cover change, population and industrial growth on river water quality are assessed with a coupled hydrological-water quality simulation model and the predominant factor contributing to pollution is identified. Also, the future water quality is projected for mid 21st century considering climate change, land use projections, population and industrial growth, and the proposed treatment for the stretch considered using socio-environmental scenarios. The effectiveness of the proposed treatment to offset the reduction in water quality from anthropogenic forcings is also assessed. The climate change effect is found to have a larger effect on water quality than other drivers, with a percentage contribution of above 70% because of the considerable sensitivity of water quality parameters to the amount of streamflow. Climate change projections combined with socio-environmental scenarios imply that the large increase in pollution due to climate change, land use land cover, industry, and population growth cannot be controlled by the current treatment proposals for 2050 by the authorities. However, providing adequate STPs to meet the population of 2050, and allowing only domestic sewage to reach STPs can help in achieving the objective of the Ganga Action Plan in the mid-21st century.
The thesis comprises of five chapters. An introduction to the problem addressed, and the objectives of the work presented in the thesis are provided in Chapter 1. Details of the case study and analysis of the sensitivity of water quality parameters to climate change and land use with idealized future scenarios are discussed in Chapter 2. In Chapter 3, the risk assessment of low water quality, eutrophication and fish kill under changing climate and land use land cover is presented. Chapter 4 presents the analysis of the individual effects of all external forcings, including climate change, land use change, population and industrial growth. Conclusions drawn from the study are presented in Chapter 5
Epistasis Detection and Phenotype Prediction in GWAS Using Machine Learning Methods
No full textGenome-wide association studies (GWAS) are used to find the association between genetic variants, Single Nucleotide Polymorphisms (SNPs), and phenotypic traits or diseases in a population. The number of GWAS has increased exponentially over the past decade due to the availability of ample data owing to the advancements in sequencing technologies. This has led to the discovery of several SNPs associated with complex diseases, but these SNPs only explain a small fraction of the disease heritability; i.e., the proportion of observed phenotypic variation between individuals in a population that is attributable to genetic variations. In these cases, univariate studies based on the single-locus analysis test the SNPs independently for the association with a particular disease/phenotype. These approaches fail to capture the complete genetic risk of complex diseases originating from the synergistic effect of multiple genes. Moreover, there is a lack of accuracy in the disease risk prediction from the genetic information due to the limited knowledge of the genetic architecture of complex diseases. Finding the interaction among SNPs is one of the ways to find this missing heritability information to evaluate the disease risks.
In this work, we propose a methodology, Learning Epistasis for Phenotype Prediction (LEPP), to discover the SNPs and the epistatic interactions among them contributing to the disease etiology. By taking the SNPs and epistatic interactions together, we also predict the disease status of a subject. Usually, GWAS identifies hundreds of thousands of SNPs, and finding interactions in such high dimensional data is a challenging task and raises issues like the curse of dimensionality. To overcome this challenge, we begin by filtering the data using multivariate feature selection methods like ReliefF, Gradient Boosting, and Random Forests, so that interacting SNPs without marginal effects are also selected along with SNPs showing marginal effects. Subsequently, we use a combination of machine learning models like Gradient Boosting, Random Forests, and Support Vector Machines to capture the non-linear relationships between features (SNPs) that are difficult for linear models to identify. We also focus on interpretability using SHAP (SHapley Additive exPlanations) to understand the predictions and the reason(s) behind the results.
Our methods are first evaluated on simulated data where the ground truth is known a priori. The best-performing methods in both feature selection and prediction steps, ReliefF and XGBoost, respectively, are chosen for the analysis of real data. Two real GWAS datasets on breast cancer and schizophrenia are used to demonstrate the efficacy of our method (LEPP). These are controlled datasets and are accessed through the dbGaP database. Redundancy of the feature set was checked using Principal Component Analysis (PCA) and the datasets were partitioned into separate subsets using a 70:30 ratio for training and testing, respectively, to assure unbiased training. We achieve maximum accuracy of 78.34% with an Area Under the Curve (AUC) value of 0.85 for the breast cancer dataset using 1000 selected SNPs. This is better than the previously reported accuracy of 60.25% on the same dataset which uses the absolute mean difference of SNP values over cases and controls for feature selection followed by the k-Nearest Neighbor method (KNN) for prediction. The reason LEPP performs better is that ReliefF is a multivariate feature selection algorithm that outperforms the univariate algorithm used in the previous work; in addition, Gradient Tree Boosting is also known to outperform KNN for classification problems. For the schizophrenia dataset, 1500 selected SNPs yielded a maximum accuracy of 76.82% and 0.84 AUC value. This is again better than the AUC value of 0.60 previously reported on this dataset which uses weighted Genetic Risk Score (wGRS) obtained by Logistic Regression coefficients. This method fails to identify the non-linear relationship between features resulting in poorer performance than XGBoost. These results show that LEPP can explain some portion of the missing heritability in these diseases which was previously not attainable. We have found novel SNPs and interactions through our analysis which provides corroboration to the current knowledge and unravels new SNPs that may be of importance. We also perform a gene-level analysis by mapping these SNPs to their corresponding genes and then explore the ways these genes and the pathways involved may influence the diseases
Learning Invariants for Verification of Programs and Control Systems
No full textDeductive verification techniques in the style of Floyd and Hoare have the potential to give us concise, compositional, and scalable proofs of the correctness of various kinds of software systems like programs and control systems. However, the major hurdle in adopting these techniques is that the verification engineer often needs to come up with adequate program invariants (like loop invariants), which may require a lot of expertise and manual effort. This thesis tries to address this problem by attempting to automate the process of finding adequate invariants, using machine learning and other techniques.
In the first part of this thesis, we introduce a data-driven learning-based technique to automate the computation of adequate invariants for the deductive verification of various classes of programs, including recursive sequential programs, and concurrent programs. We consider standard pre-post specifications for these programs. Deductive verification of programs can be viewed as finding a solution to a system of Constrained Horn Clauses (CHCs) induced by the program and its specification. Earlier works on data-driven learning of invariants like ICE learning (Garg et al, 2014) can only handle linear CHCs. However, many deductive verification techniques like Requires-Ensures (for recursive sequential programs) and Rely-Guarantee and Owicki-Gries (for concurrent programs) induce non-linear CHCs. We propose a technique called Horn-ICE, which extends the ICE learning technique to solve non-linear CHCs, thereby allowing us to automate the deductive verification of a variety of new classes of programs.
Non-linear CHCs give rise to Horn implications counterexamples (Horn clauses) in general, so our learning setup considers a semi-labelled dataset along with its metadata in the form of Horn clauses. We have implemented a decision tree based classifier for this semi-labelled dataset. This classifier learns candidate invariants for a deductive verification task. We demonstrate the performance of this automated program verification technique using standard benchmarks like SV-COMP. On the sequential benchmark suite, our tool is on par with the state of the art tool Z3/PDR. On the recursive benchmark suite, we outperform the state of the art tool Ultimate Automizer (Heizmann et al, 2013) . On the concurrent benchmark suite, we are able to verify all of the programs within a few seconds.
In the second part of this thesis, we consider sampled-data control systems that control a continuous plant using a discrete controller, and the band convergence property for these systems. Our first contribution in this part of the thesis is a deductive verification technique for sampled-data control systems. This technique symbolically executes the system from the initial state onwards until an adequate invariant is reached. For this verification technique, we developed a concolic execution based white-box approach for learning adequate invariants. We demonstrate the performance of our automated sampled-data control system verification technique using standard Simulink models. Our toolchain is able to verify band convergence properties for most of these models within a few seconds
Stirring and mixing driven by mesoscale eddies in the stratified Bay of Bengal
No full textThe stirring of passive tracers driven by altimetry-derived daily surface geostrophic currents is studied on subseasonal timescales in the Bay of Bengal. Advection of latitudinal and longitudinal bands highlights the chaotic nature of stirring in the Bay via repeated straining and filamentation of the tracer field. An immediate finding is that stirring is local, i.e., of the scale of the eddies, and does not span the entire basin. Further, stirring rates are enhanced along the coast of the Bay and are relatively higher in the pre-and post-monsoonal seasons. The spatially non-uniform stirring at the surface of the Bay is reflected in long-tailed probability density functions of Finite-Time Lyapunov Exponents (FTLEs), which become more stretched for longer time intervals. Quantitatively, advection for a week shows that mean FTLEs lie between 0.130.07 day, while extremes reach almost 0.6 day. Averaged over the Bay, relative dispersion initially grows exponentially, followed by a power-law at scales between approximately 100 and 250 km, which finally transitions to an eddy-diffusive regime. Quantitatively, below 250 km, a scale-dependent diffusion coefficient is extracted that behaves as a power-law with cluster size, while above 250 km, eddy-diffusivities range from ms in different regions of the Bay. These estimates provide a useful guide for resolution-dependent diffusivities in numerical models that hope to properly represent surface stirring in the Bay.\\
A particularly important tracer field in the Bay is the sea surface salinity; indeed, freshwater from rivers influences Indian summer monsoon rainfall and tropical cyclones by stratifying the upper layer and warming the subsurface ocean in the Bay of Bengal. We use {\it in situ} and satellite data with reanalysis to showcase how river water experiences a significant increase in salinity on sub-seasonal timescales. This involves the trapping and homogenization of freshwater by a cyclonic eddy in the Bay. Using a specific example from 2015, river water is shown to enter an eddy along its attracting manifolds within a period of two weeks. This leads to the formation of a highly stratified subsurface layer within the eddy. When freshest, the eddy has the largest sea-level anomaly, spins fastest, and supports strong lateral gradients in salinity. Subsequently, observations reveal a progressive increase in salinity inside the eddy within a month. In particular, salty water spirals in, and freshwater is pulled out across the eddy boundary. Lagrangian experiments elucidate this process, whereby horizontal chaotic mixing provides a mechanism for the rapid increase in surface salinity.\\
The eddy-freshwater interaction, or adjustment, is then studied using a high-resolution Regional Ocean Modeling System. Apart from lateral advection, a mixed layer salinity budget shows the importance of ageostrophic vertical advection during the evolution of salinity within the eddy. An analysis of the depth-integrated eddy kinetic energy indicates the development of both barotropic and baroclinic instabilities. The vertical profile associated with these conversion terms reveals that the surface freshwater was likely involved in developing baroclinic terms in the mixed layer. In addition, an eddy available potential energy (EPE) budget suggests that the entrainment of the river water raises the EPE, which is reflected in the development of gradients in salinity within the eddy. The EPE is lowered with homogenization, signifying irreversible mixing. Further, EPE rates are modulated by the correlation of buoyancy fluxes with density anomalies, which involves lateral advection of freshwater associated with surface cooling and local, regional rainfall. Finally, the adjustment of this freshwater eddy triggers submesoscale dynamics that appear to be an integral part of salinity homogenization. The observation and reanalysis data also showcase the presence of these events across different years, thus bringing out the broader impact of mixing freshwater into high salinity ambient water by eddies in the Bay. This pathway is distinct from vertical diffusive mixing and is likely to be important for the evolution of salinity in the Bay of Bengal.Ministry of Education (MoE), Indian Institute of Science, Bengaluru; University Grants Commission; National Monsoon Mission, Indian Institute of Tropical Meteorology, Pune; Divecha Centre for Climate Change, Indian Institute of Science, Bengalur
A Novel Passive Regenerative Snubber for the Phase-Shifted Full-Bridge Converter: Analysis, Design and Experimental Verification
No full textThe development of Wide Bandgap (WBG) devices has enabled power electronic converters
to operate at much higher frequencies, voltages and high power. Working at a higher
switching frequency minimises the size of magnetics but results in significant switching
losses and electromagnetic interference (EMI) noise. Thus, it necessitates the use of soft-switching
techniques to reduce these losses. Phase-Shifted Full-Bridge (PSFB) Converter
is the most widely used soft-switching topology in the high-voltage and high-power, unidirectional,
DC-DC conversion. The phase shift PWM control utilises the converter parasitics
to achieve zero voltage switching (ZVS) turn ON. The gating technique allows the
magnetic energy stored in the leakage inductance of the isolation transformer to charge
and discharge the output capacitances of the inverter leg. However, the converter suffers
from severe voltage overshoots across the rectifier bridge during the zero to the active state
transition. The resonant circuit formed between the transformer leakage inductance and
the parasitic diode capacitance of the rectifier is responsible for the high-voltage ringing.
Many passive and active snubbers are presented in the literature to mitigate the high-voltage
overshoots across the diode bridge. While passive snubbers are relatively simple
to implement than active snubbers, they are lossy. On the other hand, the active snubbers
require additional gate driver circuitry and complex control.
The first part of the thesis proposes a novel passive regenerative snubber to overcome
the mentioned drawbacks of the existing snubbers. The proposed snubber is ideally lossless
with no control complexity. The work covers a detailed analysis of the PSFB operation
with the proposed snubber while obtaining closed-form expressions for the converter state
variables at the end of each topological stage. The study considers all the major converter
parasitics, such as transformer leakage and magnetising inductances, and parasitic capacitances
of the converter. Given the new snubber, the thesis also lays out a step-by-step
PSFB design procedure utilising the analysis carried out in the first part of the work. The
design aimed to develop a 100 kHz PSFB for an input voltage of 360-440 V in the output
power range of 0.5-1.5 kW at a fixed output voltage of 48 V. The design approach focuses
on two design objectives - All inverter switches must achieve ZVS turn ON and the
desired converter gain for all possible operating conditions.
A hardware prototype is built and tested. The experimental results validate the effectiveness
of the snubber in reducing the voltage overshoot. Further, the analysis and design
accuracy is verified using the measured state variables. The work, at last, presents the
overall converter efficiency and the loss distribution among the converter components
Efficient Constructions of Streaming Codes
No full textStreaming codes are a class of erasure codes that operate on a stream of packets and enable recovery of dropped or erased packets under a decoding-delay constraint. The primary focus of this thesis is on providing constructions and performance bounds for streaming codes in some settings of practical interest. A secondary focus of this thesis is on coded distributed computation. In the streaming code literature, a sliding window (SW) channel model is often adopted, under which, within any SW of fixed time duration, the channel is permitted to introduce either a single erasure burst or, else, a set of arbitrary erasures. Erasure patterns conforming to this constraint are termed admissible erasure patterns. A streaming code operating on this channel is required to recover from all admissible erasure patterns and, furthermore, do so under a decoding delay constraint. Prior rate-optimal constructions of such streaming codes were mostly based on the diagonal embedding (DE) of codewords of a scalar block code within the packet stream and required a field size that was quadratic in the delay parameter. This thesis introduces a variant of DE called staggered diagonal embedding (SDE), in which codewords of scalar block code are embedded diagonally with carefully chosen gaps in the packet stream. The SDE approach helps to reduce the impact of burst erasures on the embedded scalar block code and, thereby, often permits the construction of rate-optimal streaming codes having linear field size. The maximum possible rate of streaming codes employing a generalization of SDE and that are based on the simple-to-implement class of MDS codes is also determined. With respect to the DE framework, a novel rate-optimal streaming code construction is presented, that covers all possible parameters and where the field size, though still quadratic, is the smallest among all currently-known explicit and general rate-optimal constructions. In a different setting, the thesis examines streaming codes where the SW channel model permits only arbitrary erasures to take place within each SW, and where this time, the goal is the minimization of the average decoding delay. Borrowing from principles of codes for distributed storage, the notion of locally recoverable streaming codes is introduced, which not only operates under the usual decoding delay constraint, it guarantees, in addition, a significantly smaller decoding delay in the more frequently-occurring case of single-packet erasure. The maximum possible rate of locally recoverable streaming codes is determined in this thesis by presenting rate-optimal constructions for all possible parameters. With the same general aim of minimizing average decoding delay, but in a slightly different direction, streaming codes operating over a symbol-erasure channel and that are optimal with respect to a metric called information-debt are investigated. A single path from source to destination will be unable to guarantee reliable erasure recovery in the presence of events such as deep fades. With this in mind, the thesis deals with multipath streaming codes, i.e., streaming codes designed for the setting when there are multiple source-destination paths, as is the case with, for example, 5G dual connectivity. Leveraging the techniques developed for single-path streaming codes, constructions and rate-bounds are derived here for the multipath setting.
The thesis also contains results on coded distributed computation. With the amount of data being generated increasing rapidly, large-scale distributed computing has become very relevant. In this thesis, two different coded distributed computation settings are investigated, namely coded MapReduce and secure distributed matrix multiplication. Coded MapReduce is a technique that trades an increase in computation for a reduced communication load, to achieve a speedup in settings where communication is the bottleneck. The placement-delivery-array is a structure initially defined in the context of coded caching to develop schemes with smaller sub-packetization. Motivated by this, a method is presented to construct coded MapReduce schemes, requiring a small number of subfiles, based on a slightly redefined version of placement-delivery-array. For the coded MapReduce framework where each Reduce function is computed by more than one node, an optimal coded shuffling scheme based entirely on XOR operations is presented, whereas the previously known scheme required operations over a larger finite field. The problem of constructing polynomial codes for secure distributed matrix multiplication can be reduced to a combinatorial problem of filling up the entries of a table subject to certain restrictions. Such tables are called degree tables, and degree tables outperforming the best-known degree tables for certain parameter ranges are constructed
Higher-Order Assembly of Protein Protected Gold Nanoclusters using Supramolecular Host-Guest Chemistry: A 40% Absolute Fluorescence Quantum Yield
No full textFor the last few decades, cancer has been one of the major public health concerns, and the mortality rate has become significant worldwide. However, because of the blessing of science, the early detection of tumors and the survival rate of patients have improved extra 5-years. Therefore, early-stage cancer detection and identification are critical in cancer treatment. Although priority must be given to providing the treatment option, simultaneously, it is also important to design new and improve the existing methods, which are the noble job for the scientist.
Recently, gold nanoclusters (Au NCs) have emerged as a promising detection approach in biomedical imaging as it exhibits molecular like properties and good photostability. Metal nanoclusters are a unique class of ultra-small particles consisting of a few to hundreds of atoms. They feature a metal core of size ˂2 nm, which is formed due to controlled aggregation. And because of the confinement of the electrons, it exhibits optoelectronic properties such as fluorescence. In this era, scientists have reported small molecules protected gold nanoclusters, but due to lack of biocompatibility and low quantum yield, it suffers from real applications such as cell imaging etc. So, designing a biocompatible highly fluorescent gold nanocluster is a challenging task to material and synthetic chemistry. Recently, Ying et al. reported protein-protected gold nanoclusters of BSA protein, which exhibit a low quantum yield (QY~6%). To address this problem, Maity et al. have reported a highly fluorescent higher-order assembly of protein-protected gold nanoclusters using supramolecular host-guest chemistry, which exhibits 40% absolute quantum yield. Towards the end, we observed a reversible aggregation-disaggregation in the presence of cucurbit[7]uril (CB7) and adamantly amine (ADA), respectively, which is important from the fundamental point of view, where in general, protein aggregation is irreversible in nature.
References:
1. Wan et al. Analyst 2020, 145, 348
2. Pradeep et al. J. Phys. Chem. C 2019, 123, 28969
3. Ying et al. J. Am. Chem. Soc. 2009, 131, 888
4. Maity et al. Nanoscale Adv. 2022, DOI: 10.1039/D2NA00123CPrime Minister's Research Fellowship (PMRF
Microscopic Analysis of Self Healing Circuits Using Image Processing
No full textOpen circuit faults are common circuit failure mechanism in Thin Film Transistor (TFT)
integrated circuits or Printed Circuit Boards (PCBs). Thin film Transistors are widely
used in flexible electronics and are manufactured using roll-to-roll methods for
application in flexible displays and image sensors, energy harvesters and wearable
electronics. Circuits and systems on flexible substrates experience open circuit failures
due to mechanical causes such as bending and stretching and electrical causes such as
electro-static discharge. It is therefore important to address the problem of open circuit faults. The above problem has been conventionally addressed by the use of new
interconnect geometries and stretchable materials. However, these are passive methods and do not solve the problem for non-mechanical causes of open faults. Another approach has been the self-healing of interconnects using a dispersion of conductive particles in an insulating medium. This dispersion is packaged over the interconnect. When a current carrying interconnect experiences and open-fault, the conductive particles of the dispersion are polarized and experience dipole-dipole attractive forces. This eventually leads to the particles chaining up to form a bridge that heals the fault. So far, the models are based on the macroscopic or system level behavior of the dispersion in response to an electric field. These models assume that there are two main forces at play – the dipole-dipole attractive force aiding the healing, and the viscous drag in the fluid inhibiting the motion of particles. In this work, we perform a microscopic analysis of each particle using image processing techniques. The image processing technique used is a robust pixel wise classification algorithm and a convolutional auto-encoder based image segmentation algorithm for particle segmentation. Essentially, the motion of each particle is tracked and the force versus inter-particle distance profile is obtained. This indicates the kind of forces at play. Experiments indicate the force roughly
varies as the inverse fourth power of distance thereby corroborating with the model of dipole-dipole interaction