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
Stochastic Optimization And Its Application In Reinforcement Learning
No full textNumerous engineering fields, such as transportation systems, manufacturing, communication networks, healthcare, and finance, frequently encounter problems requiring optimization in the presence of uncertainty. Simulation-based optimization is a workable substitute for accurate analytical solutions because of the numerous input variables and the need for a system model. Smoothed functional (SF) algorithms
belong to the class of simultaneous perturbation methods that have been found useful for stochastic optimization problems, particularly in high-dimensional parameter spaces. SF methods update the gradient of the objective using function measurements involving parameters
that are perturbed simultaneously along all component directions. \cite{katkul} originally developed the SF gradient procedure. This results in the objective function
getting smoothed because of the convolution. The objective function smoothing
that results from the convolution with a smoothing density function can help the algorithm converge to a global minimum or a point close to it.
First, we present a stochastic gradient algorithm for minimizing a smooth objective function that is an expectation over
noisy cost samples and only the latter are observed for any given parameter. Our algorithm employs a gradient estimation scheme with random perturbations, which are formed using the truncated Cauchy distribution from the sphere. We analyze the bias and variance of the proposed gradient estimator. Our algorithm is found to be particularly useful in the case when the objective function is non-convex and the parameter dimension is high. From an asymptotic convergence analysis, we establish that our algorithm converges almost surely to the set of stationary points of the objective function and obtains the asymptotic convergence rate. We also show that our algorithm avoids unstable equilibria, implying convergence to local minima. Further, we perform a non-asymptotic convergence analysis of our algorithm. In particular, we establish here a non-asymptotic bound for finding an -stationary point of the non-convex objective function. Finally, we demonstrate numerically through simulations that our algorithm outperforms GSF, SPSA, and RDSA by a significant margin over a few non-convex settings, and we further validate its performance over convex (noisy) objectives.
Next, we consider the problem of control in the setting of reinforcement learning (RL), where model information is not available. Policy gradient algorithms are a popular solution approach for this problem and are usually shown to converge to a stationary point of the value function. We propose two policy Newton algorithms that incorporate cubic regularization. Both algorithms employ the likelihood ratio method to form estimates of the gradient and Hessian of the value function using sample trajectories. The first algorithm requires an exact solution of the cubic regularized problem in each iteration, while the second algorithm employs an efficient gradient descent-based approximation to the cubic regularized problem. We establish convergence of our proposed algorithms to a second-order stationary point (SOSP) of the value function, which results in the avoidance of traps in the form of saddle points. In particular, the sample complexity of our algorithms towards finding an -SOSP is , and this is a significant improvement over the previous state-of-the-art sample complexity of
Differential DNA methylation patterns in primary and recurrent glioblastomas : Protumorigenic roles of Tumor Endothelial Marker 8 (TEM8) and ASAP1 genes
No full textGlioblastomas (GBMs) are clinically aggressive malignant brain tumors in adults with a median survival of 14 months on diagnosis. Therapeutic interventions such as debulking surgeries, chemotherapy, and ionizing radiation, while relieving symptomatic disease, are not useful in preventing subsequent relapse (or recurrence). Recurrences are virtually inevitable in glioblastoma, due to the challenges in total surgical resection and/or tumors being highly invasive. Recurrent tumors are chemoradioresistant, and prove to be fatal in most cases.
Several lines of evidence point to the fact that both Temozolomide (the chemotherapeutic drug of choice) and ionizing radiation, can bring about stable changes in glioblastoma cells. As glioblastoma cells exhibit inherent plasticity, we hypothesized that there exists DNA methylation-based epigenetic changes in therapy-exposed recurrent tumors compared to treatment-naïve tumors in patient-matched pairs. To answer this, we compared the DNA methylation profiles of 11 primary GBMs (first occurrences) and 13 recurrent GBMs, using the Illumina 450K BeadChip Array, and identified 1225 hypermethylated regions and 526 hypomethylated regions in recurrent tumors compared to their primary counterpart. We mapped the probes to gene promoters and were able to identify promoter hypermethylation of several tumor suppressor genes, and hypomethylation of several genes involved in cancer progression.
We validated transcriptional dysregulation for some of these genes by expression and immunohistochemical analysis; and functionally characterized two of them: Tumor Endothelial Marker 8 (TEM8) and Arf-GAP with SH3, Ankyrin and PH domain protein 1 (ASAP1).
In GBMs, the role and function of TEM8 remain unknown. Using immunohistochemistry, we found that while TEM8 expression is non-detectable in control brain tissues, it is elevated in some lower-grade gliomas and in all GBM tumors. In a seperate retrospective cohort of 30 pairs of patient-matched primary and recurrent glioblastomas, increased expression of TEM8 was found in recurrent tumors. In silico analyses revealed TEM8 overexpression in glioblastoma patients is related to a worse prognosis. In-vitro, using overexpression and knockdown approaches, we found that TEM8 expression conferred proliferation advantage, invasive and migratory properties in glioma cells apart from chemo- and radioresistance, and stemness. Mechanistically, we demonstrated that TEM8 expression stabilized β-catenin levels and its’ target gene expression in glioblastoma cells. Using small molecule inhibitors, we demonstrated that TEM8 activates β-catenin via Src/PI3K/GSK3β pathway in glioblastomas.
Likewise, we explored the role and function of the ASAP1 gene in glioblastomas. We demonstrated that ASAP1 expression in GBM confers poorer survival and is upregulated in recurrent tumors. Knockdown of ASAP1 resulted in reduced proliferation, migration, and reduced cell size in glioblastoma cells. We observed that ASAP1 knockdown leads to impaired mTOR and ERK signaling, especially in response to growth factors such as EGF. We, therefore, demonstrate that ASAP1 is a key player in integrating and transmitting receptor tyrosine kinase signaling to mediate growth in glioblastoma cells. Together these studies suggest that preferential upregulation of protumorigenic genes by promoter hypomethylation may aid the aggressiveness of recurrent GBM tumors
Experimental and Theoretical Investigations on High Voltage Polymeric Insulators
No full textHigh Voltage Ceramic and glass Insulators have been widely used by various transmission and
distribution utilities for several decades across the globe. Recently composite or silicone rubber
insulators have evolved and are now replacing ceramic/glass insulators due to their improved
advantages; however, these Insulators suffer from degradation over a period of service. The
first few chapters of the thesis deal with the study of silicon rubber/polymer insulators under
various climatic conditions. Exhaustive experimental studies were conducted to understand
the degradation of insulators under different climatic conditions which prevail in the Country.
Studies on polymer insulators under sub-zero and under extremely high-temperature conditions
were attempted experimentally to evaluate their performance. During experimentation,
the leakage current was continuously monitored. Later, material analysis, which is a very
important aspect and essential to correlate with the morphological changes of the insulator
surface, was examined. The experimental investigations demonstrate that there is a need to
conduct multi-stress experimentation under specific climatic conditions before the Insulators
are installed in the field. The next portion of the thesis work deals with the failure mechanism
of a Fibre Reinforced Plastic (FRP) Rod. Some portion of the work deals with mathematical
analysis being extended to condition monitoring of dielectric surfaces and understanding the
performance of FRP rods under high AC voltages. Further, experimental investigations are
performed on FRP Rods to analyze the behaviour witnessed, as the field failures reported on
Silicon rubber Insulators, interesting results are reported. Condition monitoring of dielectric
surfaces is very important; hence it was felt necessary to analyze the field performance of transmission/
distribution composite Insulators. To understand further, a mathematical analysis
based on Chaos has been evaluated for leakage current data and quantization of comparative
degradation for a dielectric surface is presented. Later, Empirical Mode Decomposition is also
used for understanding leakage current and implied degradation under minimal data conditions,
and the results are analyzed and presented. Subsequently, the Surface electric field of insulators
exposed to HVDC is studied considering the temporal boundary conditions which may arise
due to the capacitive-resistive transients. The last portion of the thesis deals with a theoretical
study of the bulk conductivity of polymer material. The Electric Field dependence of conductivity
on the application of voltage and subsequent space charge distribution is attempted, and
the results are analyzed and presented. In short, this thesis is a work where both experimental,
simulation and theoretical studies pertaining to silicone rubber insulators are presented
Optimal Placement and Traffic Steering of VNFs and Edge Servers using Column Generation in Data Center Networks
No full textTelecom Service Providers (TSPs) were traditionally dependent on physical devices to provide end-to-end communication. The services provided were high quality and stable but low in agility and hardware-dependent. As the demand for quick deployment of diverse services increased, TSP-s needed much higher flexibility and agility. This is how Network Functions Virtualization (NFV) came into being. NFV is the concept of replacing dedicated hardware with commercial-off-the-shelf (COTS) servers. It decouples the physical hardware and the function running on it. A network function can be dispatched as an instance of the software, called a Virtual Network Function (VNF). Thus, a service can be decomposed into several VNFs that can be run on industry-standard physical servers. The optimal placement of these VNFs is a potential question for TSPs to reduce the overall cost.
We first study a network operations problem where we optimally deploy VNFs in Service Chains (SCs) such that the maximum consumed bandwidth across network links is minimized. The network parameters (link bandwidths, compute capacities of nodes, link propagation delays, etc.) and the number of SCs are known a priori. The problem formulated is a large Mixed-Integer Linear Program (MILP). We use the Column Generation (CG) technique to solve the problem optimally. Through various examples, we show the power of CG. We compare our results with recent heuristics and demonstrate that our approach performs better as it gives exact optimal solutions quickly.
Second, we extend our previous setup to the online case where the number of SCs is not known a priori. We serve SC requests as they come. A new SC is implemented on the "residual network" while the previously deployed SCs are undisturbed. The problem formulated is a large MILP, and we use CG as the solution technique. The results show the percentage improvement in the solutions over those obtained using heuristics.
Next, we study a network design problem in an Edge Computing Environment. A general communication network has a single Data Center (DC) in its "core," which serves as a gateway to the Internet. For delay-constrained services of the kind needed by online gaming, this model does not suffice because the propagation delay between the subscriber and the DC may be too high. This requires some servers to be located close to the network edge. Thus, the question of the optimal placement of these edge servers arises. To lower the network design cost, it is also essential to ensure good traffic routing, so that aggregate traffic on each link remains as low as possible. This enables lower capacity assignment on each link and thereby minimizes design cost.
We study a novel joint optimization problem of network design cost minimization. Edge server placement cost and link capacity assignment cost constitute the total cost. The problem formulated is a large MILP, and we again use CG to solve it. We compare our results with many heuristics and show the improvement in design cost.
Finally, we extend the above work by relaxing some assumptions and constraints. Unlike previously, we consider servers with different capacities. Also, a server can serve more than one request depending on its core capabilities. We also consider the split-and-merge of an SC through various paths in the network. The formulated problem can also provide the minimum number of servers to be used. Again, the formulation is a large MILP, and CG is used to solve it exactly
Understanding the architecture of mitochondrial presequence translocase machinery and its implications in ALS progression
No full textThe current thesis describes the structural organization of the presequence translocase machinery and its functional association with ALS pathogenesis. The TIM23 complex is essential for the mitochondrial biogenesis of polypeptides with N-terminus targeting sequence into the matrix and inner membrane. Hence any defects in the import machinery are known to cause mitochondrial dysfunction affecting cellular homeostasis. The importance of the presequence translocase machinery can be highlighted as its architecture, and overall import mechanisms have remained evolutionary conserved. However, because of the complex nature of a multicellular organism, the human TIM23 complex was proposed to have multiple forms distinguished by their role in housekeeping function and disease association. The findings of the current thesis provide insights into the structural dynamics of the human TIM23 machinery and how variants of an import motor complex protein differentially influence the matrix protein translocation. Additionally, the work focuses on the functional characterization of the presequence translocase machinery under a disease background, wherein the binding of mutant results in the altered translocation of a substrate leading to mitochondrial dysfunction.CSI
Device-Circuit Reliability Co-Design in High voltage and Power devices
No full textFor the last four decades, silicon CMOS technology has captured a significant share in IC, smart power IC, SoC, and the power device market. But there is aggressive research on other materials such as graphene & similar 2D materials and wideband gap materials. But, several aspects, including the fabrication process to improve device performance [7,9,12,13], understanding the device reliability physics [8,10,11], interconnection and packaging, need to be matured before these compound materials take the limelight. Besides these, a fab set-up for large scale production requires high NRE capital. On the other hand, Silicon had been through a great degree of maturity. Moreover, for intelligent power applications, Silicon has superior reliability. Therefore, Silicon is predicted to capture the power device market till the other materials gain perfection.
Besides the commercial market, the requirement for discrete and integrated power device technology within the strategic sector is enormous. Discrete power switching devices (often power MOS or IGBT switch) and power RF devices are used in numerous onboard power electronic and power RF applications. The first part of this work strives to bridge the device-circuit co-design gap that has severely limited predictive modelling of circuits for high power applications such as Radio Frequency Power Amplifiers (RF PAs) using high power devices LDMOSs and GaN HEMTs. A correlation between the device’s intrinsic parameters and PA performance is explored, and the iterative process aims to provide a high-performing circuit.
LDMOS is one of the prominent power devices which adopts CMOS processing and easy integration. The lateral double diffused MOSFET (LDMOS) is the predominant power device in implementing Power integrated circuits PICs because of its attractive electrical characteristics such as low on-resistance, high breakdown voltage, high input impedance and fast switching frequency. To obtain high breakdown voltage with low on-resistance for LDMOS, RESURF technology is used, in which the vertical p-n junction depletion layer between the n-type drift region and p-type epitaxial region and its interaction with lateral p-n junction depletion between the p-type channel and n-type drift region is optimised to reduce the surface electric field to obtain high breakdown voltage. A field plate (FP) on top of the gate and a drain field plate on top of the drift region is introduced to improve breakdown voltage further. These field plates help reduce feedback capacitance (Cgd) and increase the breakdown voltage FP LDMOS.
This work explores optimising the field plates for achieving breakdown characteristics above 900V without altering the on-resistance of the devices [2]. It covers major classical power devices from conventional design to non-conventional device designs. Conventional devices without field plates show 30% lower breakdown voltages than those with field plates. It can be concluded that field plates play a vital role in enhancing the breakdown characteristics of the device. Taking optimised design further, the field plates are introduced into non-conventional devices, where RESURF and SOI-based devices are explored. Along with the performance studies, the reliability of these structures is also explored. Regarding reliability, RESURF based devices show a higher tendency of deviation/degradation when stressed, up to 20% higher than the breakdown condition like that of the conventional devices without Fp. Further, the role of each field plate individually under the ESD condition is explored [3]. It was clearly understood that field plates at the source side play a significant role in distributing the junction electric field, while field relaxation at the drain side helps in improving the failure threshold. Gate and Source field plates improve the trigger voltage characteristics up to 54%, while drain field plate improves the failure threshold up to 60%.
Power semiconductor device industries are aggressively looking for system-on-chip (SOC) solutions for power amplifier (PA) circuits and are exploring different technologies, including gallium arsenide (GaAs), to more recent and intriguing gallium nitride (GaN) technologies, for power transistors with Radio Frequency (RF) applications. The advantage of silicon technology for RF applications lies mature fabrication process at low cost and their easy integration capability with the CMOS technology. Applications that require radio frequency power amplifiers such as broadcast, ISM (industrial, scientific, and medical), avionics, radar, wideband communications, telecom & satellite communications, RF heating applications, etc. Among them, the 50V RF LDMOS device is mainly used in wireless broadcast, ISM, and radar, which requires a higher breakdown voltage and power density. For 50V RF LDMOS with 0.35µm CMOS technology, the breakdown voltage must be higher than 100 V to guarantee a reliable operation. RESURF technology and the introduction of field plates have improved the breakdown voltage to 114V. Both DC and RF performance of various designs of FP RF LDMOS are evaluated, and results show that industry-leading performance is achieved [1]. Electrostatic Discharge (ESD) robustness was studied for the FP RF LDMOS designs as ESD has been identified as a source of damage to unprotected devices. Hot Carrier Injection (HCI) reliability was also investigated to address the complete reliability of these devices.
Though Silicon-based transistors have the advantage of mature technology, the requirement for high power and high-frequency devices demands transistors based on semiconductor materials with large breakdown voltage and high electron velocity. GaN is an attractive candidate for power amplifier applications because of several superior qualities in amplifier applications achieved due to its semiconductor properties. The wide bandgap in Gallium Nitride based transistors results in higher breakdown voltages because the ultimate breakdown field is the field required for band-to-band impact ionisation. Also, its high electron saturation velocities allow high-frequency operation. GaN can be used to fabricate high electron mobility transistors (HEMTs), which have high carrier concentration and higher electron mobility due to reduced ionised impurity scattering. A rigorous device-circuit co-design investigation of AlN/GaN HEMT to explore its feasibility for power amplifier operation at frequencies > 1THz. Both class A and class AB operations were invested. A novel device-circuit co-design methodology was adopted [4], which involves (i) device design optimisation using a well-calibrated TCAD setup, (ii) careful extraction of large-signal model cards with I-V, C-V & S-parameter matching, and finally (iii) source-load pull-based power amplifier design/exploration, for every device design investigated. For PA operation, both class A and class AB operations were invested while exploring PA gain, output power, efficiency at 1dB compression point, and linearity through dual-tone (IMD3) investigations. Besides, a complete range of device design parameters was investigated to explore the ultimate scalability limit and narrow down the device design window that can enable THz operation.
The last two decades have witnessed significant scaling in MOS technology from sub-micron to sub-nm level. To achieve devices with good performance at a small dimension, it was essential to explore new device architectures which could offer subthreshold swing (SS) values below 60 mV/dec. Several CMOS-like structures were designed, such as Fin-FETs, Nanowire Gate All Around (GAA) MOSFETs, Carbon nanotube FETs, Tunnel FETs, etc., which could lower the leakage current at small dimensions. Among these, TFET could achieve SS less than 60 mV/dec attributed to a fundamentally different mechanism for carrier injection. Hence, TFET is considered a future on the roadmap. A prediction of reliability is essential in choosing a device for a particular application. Therefore, it is necessary to understand the reliability of devices at the design stage itself. As technologies advance towards the deep submicron, the ESD (Electrostatic Discharge) protection design issues have become more critical. The second part of this work tries to understand the ESD robustness of a couple of novel Tunnel FET architectures.
A novel Fin-enabled vertical or area-scaled tunnelling FET is proposed for sub- 10-nm channel length operation. This device enables a smooth transition from FinFET technology to Fin-based vertical TFETs, while enjoying the benefits of FinFET architecture. To make this device commercial, it’s essential to understand the reliability performance of this device. This work explores the reliability physics of this device with detailed physical insight into the device’s operation and failure under ESD stress conditions [5]. The proposed device has a deep N+ implant underneath the P+ source, like adding a pocket between the source and gate for the ESD protection applications. Early avalanche assisted BTBT at the source-pocket junction, in addition to the drain-substrate junction, causes the device to turn on at lower voltages, lower self-heating resulting in improved failure current in the proposed device with less area overhead
Complexity and Entanglement: From quantum gravity to many-body systems
No full textIn recent years, complexity and entanglement have emerged as two fundamental computational measures and played a significant role in shaping our understanding of various phenomena, from the geometric nature of quantum gravity to the critical phenomena in many-body systems. In the first part of the thesis, we primarily focus on complexity in three different aspects. We modify Nielsen’s original arguments of traditional quantum gate counting, utilizing higher-order integrators of the Suzuki-Trotter method. This provides a volume-law scaling of complexity that is consistent with holographic proposals. Then we discuss the higher-dimension generalization of path integral complexity and its holographic interpretation using the AdS/BCFT correspondence. Later, we turn our attention to subregion complexity, which is a version of complexity that plays a significant role in understanding the black hole information problem. In the second part of the thesis, we zoom in to the entanglement for both pure states and mixed states. First, we discuss the capacity of entanglement in diverse scenarios, from operator excitations in quantum field theory to the phenomena of quantum chaos in many-body systems. We then delve into the details of the mixed state entanglement, introducing a measure known as the balance partial entanglement. In several examples, we show that it generalizes the reflected entropy and equals the entanglement wedge cross-section from the gravity perspective.University Grants Commissio
Multi-linear Disassembly Path Determination: A Geometric Approach
No full textThis thesis presents a geometric approach for determining the orientation-preserving disassembly paths of polyhedral
assembly components. Exact disassembly path determination of the components is essential because the reversal of the disassembly paths provides the paths for the components to assemble into a functional product. Single straight-line paths for disassembly algorithms are available in the literature. Multilinear disassembly requires the determination of the exact Minkowski sum. Minkowski sum approaches, which are related to configuration space, have been used in path planning in both robotics and assembly. But they fail when the assembly components have mating boundaries. This limitation arises due to the failure to capture the contact spaces between the mating boundaries in the Minkowski sum. We have used non-regularized Boolean to capture these contact spaces in the form of lower-dimensional features, which are usually eliminated in regularized Boolean. These lower-dimensional features are characterized into different path elements, which provide the local motion space for a component to move in configuration space. The composition of these path elements models the disassembly paths. To accomplish this board goal, a few sub-problems have been solved. The Minkowski sum of a pair of arbitrary solids requires both Boolean and convex decomposition, which is then used to determine the disassembly paths for a component of an assembly. To achieve the broad objective of the thesis, the following contributions are made:
(a) An algorithm has been developed for Boolean of a large number of polyhedral solids. It is based on cell classification
without a priori point classification using Slice representation. Contact spaces are accurately captured as lower dimensional features, which is the requirement of the present problem, making it a non-regularized Boolean.
Although the method obviates the need for complete boundary evaluation, it can provide exact point classification,
which is as accurate as B-rep and as fast as voxel representation of solids.
(b) The slice representation not only enables easy multi-Boolean. It also enables a “core and crust" model to partition
a tessellated solid with an arbitrary topology into a set of disjoint convex pieces. The core comprises a set of
prismatic solids of identical square sections contained in the solid and represents an approximate convex
decomposition (ACD). The crust comprises a set of convex solids of arbitrary form and supplements the (ACD)
to make it exact. It is fast and robust to handle defective solids such as solids with missing patches and self intersections. It also provides a unique capability of selective convex decomposition of any specific domain of
interest.
(c) Efficient and exact union of the hundreds of pairwise Minkowski sums of the combination of the convex
components is enabled through the slice representation without the loss of the essential lower-dimensional
features. A graph of the available motion space in individual grid-cells in the slice representation is then analysed
to construct all the paths with heterogeneous degrees of freedom. i.e. each disassembly path is multilinear,
multiway and multi-dimensional.
(d) The method developed for disassembling two components is shown to be general enough for analysis of assemblies
with an arbitrary number of components where each target sub-assembly and its complement are treated as the
two components
Growth of hole-conducting Cu2O & CuO semiconductors using CVD for next-generation electronic devices: Thin Films Transistors, Memristors, and Gas Sensors
No full textSemiconducting oxides with visible-range transparency and high electrical conductivity have tremendous potential for transparent CMOS devices. Oxide semiconducting materials are a good choice for emerging transparent electronics due to their stability, good transparency, wide bandgap, and low processing temperature. However, oxide semiconductors arelagging in electrical properties compared to single-crystal silicon. Still, they give excellent competition to amorphous silicon with low cost of production and eco-friendly nature.
Transparent electronic applications are limited by the lack of availability of p-type oxide semiconductors with adequate performance. In digital electronics, a p-type oxide transistor is a key component of CMOS devices, but due to the unavailability of a p-type oxide transistor semiconductor with performance comparable to n-type, it's challenging to develop a high-performance transparent CMOS device. So, p-type oxides are the primary culprit and bottleneck in achieving high performance in their devices. Although much effort has been put into P-N junction devices such as solar cells, LEDs, and CMOS electronics, their performance is still limited. The only applications that can be accomplished to date are based on n-type oxide semiconductors. If high-performance p-type oxide semiconductors can be synthesized, it will usher in a new age of next-generation transparent electronics devices that will impact many aspects of our everyday life.
The main reason for this discrepancy lies in the difference in the low mobility of holes in p-type oxide semiconductors compared to the high electron mobility of n-type oxide semiconductors, as the effective mass of electrons is lower than the effective mass of holes. There is a vast list of n-type oxide semiconductors with low effective mass; however, there are relatively few p-type oxide semiconductor materials. Unfortunately, none have a comparable effective mass.
Cu2O is a rare transition-metal oxide with a bandgap of 2.2 eV and one of the few oxides that show p-type conductivity with high Hall mobility. Unlike other p-type semiconductor metal-oxides, Cu2O has the high hole mobility needed for transparent electronics. Unfortunately, the thin-film deposition of pure Cu2O is not trivial, especially with physical vapour deposition (PVD). Pure phase Cu2O is formed in a narrow pressure-temperature window, only under precise oxygen potential. Therefore, we have deposited Cu2O using CVD. For device-grade films, chemical vapour deposition (CVD) is superior, as it allows a more robust control of deposition parameters, leading to better uniformity, topology control, and step coverage over large areas. CVD also gives the freedom to control the supersaturation, so crystallinity, morphology, and grain size can be engineered, which plays a significant role in device performance.
As previous literature states, theoretically, Cu2O has the potential to show good mobility, but unfortunately, the performance of the reported device is neither remarkable nor consistence. Therefore, to achieve a better performance in this work, we fabricated the TFTs using CVD-grown Cu2O with high Hall mobility on four dielectrics. We have also investigated the origin of poor device characteristics in conventional reposted Cu2O-TFTs. We have also proposed a systematic approach to passivating these interface traps, improving the field-effect mobility, subthreshold swing, threshold voltage, and enhanced gate-bias-voltage stressing stability.
The bottleneck of efficient implementation of CMOS data handing is challenging due to the ‘memory wall’. A memristor with tunable resistance is an ideal building block for storing memory. As a memristor is an emerging fourth electronic element, lots of work must be done in the material process engineering domain. Here, we have proposed CVD deposited resistive switching layers memristor. In this work, the device stack of the memristor contains an intrinsic defective layer of cupric oxide (CuO) and cuprous oxide (Cu2O) sandwiched between electrodes. These Cu2O and CuO layers were deposited at four different temperatures (300C, 400C, 500C, and 600C). Electrical and material characterizations illustrate that grains or corresponding grain boundaries play a vital role in controlling the switching behavior. Overall, we demonstrate good consistency in device parameters such as reproducibility, endurance, and retention data for the film deposited at 600C.
Further, several micro and nanostructures of Cu2O have been utilized in gas sensing of oxidizing and reducing gases. However, large-area Cu2O films are needed for the mass production of sensors, which was a challenge. In this work, we also report a chemiresistive gas sensor based on pure-phase Cu2O deposited by chemical vapour deposition (CVD). The sensing results of Cu2O films have been explained from the standpoint of roughness, morphology, and unpassivated bonds present on the surface of films. At an operating temperature of 200℃, the sensor is highly sensitive to ammonia. The device's response time (response) and recovery time (recovery) were found to be decent for practical applications. Unlike competing techniques for Cu2O deposition, Cu2O from chemical vapour deposition leads to more repeatable, stable, and reproducible sensors.DS
Multi-vehicle anticipation-based models for describing driver behaviour in heterogeneous and disorderly traffic conditions
No full textDriver behaviour models are widely used in the traffic engineering literature and practice. They are used for understanding drivers’ manoeuvring decisions in traffic streams. They also form the building blocks of microscopic traffic simulation tools, which are employed for traffic flow analysis and capacity estimation necessary for the design and operation of traffic facilities and evaluation of operational strategies. Most driver behaviour models in the literature assume homogeneous and orderly traffic conditions, characterised by homogeneity (i.e., only passenger cars comprise the traffic streams) and orderliness (i.e., vehicles move only in the longitudinal direction, except when changing lanes). Models developed with such assumptions cannot be applied to analyse heterogeneous, disorderly (HD) traffic conditions. This is because HD traffic streams, unlike homogeneous traffic streams, comprise a wide variety of vehicle classes with considerably different physical and operational characteristics. Moreover, driving in HD traffic streams is characterised by weaker lane discipline due to a greater extent of lateral movements than that in homogeneous traffic streams.
This dissertation aims to formulate and apply driver behaviour models for HD traffic streams on uninterrupted traffic facilities while considering the following aspects – (1) the multi-vehicle anticipation (MVA) behaviour, where drivers’ manoeuvring decisions are influenced by multiple vehicles around them, as opposed to a single lead vehicle ahead, (2) the treatment of driver behaviour as a combination of different manoeuvring decisions, such as the decision of whether to accelerate, decelerate, or remain in same speed (represented by a discrete variable) and the decision of the extent of acceleration or deceleration (represented by continuous variables) – as opposed to a single, continuous variable representing all these facets of driver behaviour, (3) the incorporation of stochasticity due to the errors drivers make in perceiving the traffic environment, and (4) the consideration of drivers’ intentions (which are typically latent to the analyst) and two-dimensional movements of vehicles simultaneously while also incorporating MVA behaviour. Specifically, the following driver behaviour models are formulated and applied to understand driver behaviour in empirical trajectory datasets from Chennai (HD traffic) and California (homogeneous traffic):
1. The first model presented in this dissertation is an MVA-based discrete-continuous choice modelling framework to model vehicles’ longitudinal movements in HD traffic streams. In this model, driver behaviour at a given time instance is treated as a combination of (a) the driver’s choice of whether to accelerate, decelerate, or maintain a constant speed – represented by a discrete variable – and (b) the extent of acceleration or deceleration – represented by continuous variables. The discrete and continuous variables representing driver behaviour are modelled using a simultaneous econometric framework. The proposed model is used to examine driver behaviour in the HD traffic dataset from Chennai. The empirical analysis reveals the significance of the MVA effect on driver behaviour. Specifically, drivers consider the relative speeds and space gaps with respect to multiple vehicles within an influence zone around their vehicle. In addition, it is found that the influence of the traffic environment on drivers’ discrete choices (whether to accelerate, decelerate, or maintain a constant speed) is not the same as that on their choices of how much to accelerate or decelerate.
2. The second model is an extension of the above model to recognise the panel data nature of vehicle trajectory datasets typically used for estimating the parameters of driver behaviour models. This model recognises the role of vehicle- and driver-specific unobserved factors (latent to the analyst), such as aggressiveness that influence driving behaviour, and such influence persists across all observations of a vehicle. Doing so helps in reducing the confounding effects of unobserved factors when the proposed model is applied to different datasets to compare driving behaviour in different traffic streams. The panel data model is used to understand and compare longitudinal driving behaviour between the HD traffic dataset of Chennai and the homogeneous traffic dataset of California. The empirical analysis reveals the presence of MVA effect on driving behaviour in the homogeneous traffic setting, too. However, drivers in the HD traffic stream are influenced by more vehicles in their vicinity than those in the homogeneous traffic stream.
3. In the third model formulation, a mixed multinomial logit-based framework is developed to recognise stochasticity in driver behaviour models due to drivers’ errors in perceiving the traffic environment. For this model, an econometric analysis is undertaken to evaluate two different ways of specifying errors in variables in discrete choice models – additive errors and multiplicative errors. It is shown that the multiplicative specification of errors has a better behavioural basis and allows better identification of parameters representing variability due to drivers’ perception errors. An application of this model to the HD traffic dataset reveals different levels of variability due to errors in the perception of different traffic environment variables. It is found that drivers may pay greater attention to (which results in lower variability in) perceiving space gaps and relative speeds with respect to vehicles directly ahead of them than those not directly ahead.
4. The fourth and final model formulation is a two-dimensional, MVA, and multi-stimuli-based latent class framework to analyse motorcyclists’ two-dimensional movements in HD traffic streams. This formulation conjectures that drivers manage their cognitive load by dividing their driving decisions into two steps – (a) higher-level, strategic intentions (of whether to accelerate, decelerate, or maintain a constant speed and whether to steer to the left of, right of, or keep straight along the longitudinal direction), which are not fully observable from vehicle trajectories (hence latent to the analyst), and (b) lower-level, tactical decisions that can be observed in vehicle trajectories, such as the specific angle of movement and the specific extent of acceleration or deceleration executed. When applied to the HD traffic dataset of Chennai, the proposed model suggests that drivers’ higher-level intentions are more strongly influenced by the microscopic traffic environment variables than their lower-level decisions, perhaps because drivers invest a greater extent of cognitive resources for making higher-level intentions than that for lower-level decisions.
Finally, a traffic simulator is developed to simulate traffic streams using the models developed in this dissertation. The simulation experiments using this simulator demonstrate that all the microscopic driver behaviour models developed in this dissertation reflect the typically observed macroscopic properties of vehicular traffic steams