Mtech Thesis Abstract 2015

M.TECH.
THESIS ABSTRACT 2015
Sr Area Page No
No
1 Signal processing, Communications & Networks 2-6
2 Microelectronics,VlSI &Technology Display 7-8
3 RF Microwaves & Photonics 9-10
4 Power &Control 11-12
1
Signal processing, Communication & Networks
Sr Title Supervisor Page

No No
1 Video Stabilization And Object Tracking Using Affine- Gupta Sumana 13

Spatio Temporal Regularity Flow
2 Modeling Twitter Hashtag Trends Jagannatham 14
Aditya K
3 A Distributed Message Dissemination Framework for Sharma Govind 15
Emergency Response Systems over Smartphones
4 Learning Distributed Document Representations for Multi- Hegde Rajesh 16
Label Document Categorization Mahanand
5 HDR imaging with conventional cameras Venkatesh KS 17
6 People counting in high density crowds from still images Venkatesh KS 18

7 Flash crowd handling in P2P live video streaming systems Singh Y N 19
8 SBL based joint sparse channel estimation and maximum Jagannatham 20

likelihood sequence detection in OSTBC MIMO-OFDM Aditya K
systems
9 A novel time domain adaptive decision feedback Jagannatham 21
equalization technique for cyclic code shift keying Aditya K
modulated wireless systems and convergence analysis of
LMS and RLS algorithms
10 Multiple beacon based robust decentralized detection in Jagannatham 22
cooperative MIMO wireless sensor networks Aditya K
11 Interference Alignment For Linear Coherent Decentralized Jagannatham 23
Parameter Estimation In Multiple Wireless Sensor Networks Aditya K
12 Battery Management Through Real Time Pricing Rajawat Ketan 24
13 Spatiotemporal Colorization of Video Using 3D Steerable Gupta Sumana 25
Pyramids
14 Stochastic resource allocation in device to device Rajawat Ketan 26
communication
15 Calibration and mapping for soot monitoring Rajawat Ketan 27
2
16 Channel Tracking and Estimation Transmit Beamforming Rajawat Ketan 28
With Frugal Feedback using KF And Quantized-KF
Approach
17 Distributed cooperative localization via ADMM Rajawat Ketan 29
18 GLRT based target detection in monostatic MIMO RADAR Jagannatham 30

systems Aditya K
19 A novel bandwidth efficient channel estimation scheme in Jagannatham 31
OSTBC-MIMO-OFDM systems using superimposed pilots Aditya K
20 Optimal Target Detection and 3-D MUSIC based Parameter Jagannatham 32
Estimation for Bistatic MIMO Radar Systems Aditya K
21 Robust precoder design for decentralized estimation in Jagannatham 33
wireless sensor networks Aditya K
22 Analysis Of Spatially Modulated Systems For Spectrum Jagannatham 34
Sensing And Cooperative Communication Applications Aditya K
23 Performance Analysis Of Decode And Forward Protocol Jagannatham 35
Based Cooperative Communication Systems Aditya K
24 Multi-User Beamforming in Frequency-Selective MIMO Jagannatham 36
MC-CDMA systems Aditya K
25 Depth estimation from single image using texture and Gupta Sumana 37
defocus cues Venkatesh K S
26 Reversible Watermarking in Images using Difference Gupta Sumana 38
Expansion
27 Learning Dynamical Processes Over Graph Via SSDL Rajawat Ketan 39
28 Online Tracking Of Low Rank Matrices From Partial

Observations Rajawat Ketan 40
29 Optimal Encoder Design & Sensor Scheduling in Wireless Rajawat 41
Sensor Network Ketan
30 On Optimal User Pairing and Opportunistic Interference Chaturvedi A K 42

Alignment in Interference Channels
3
31 Energy Harvesting Cognitive Radio : Save-Sensing-Transmit Banerjee Adrish 43
Tradeoffs under Primary User Traffic
32 Voice Activity Detection using Deep Belief Networks by Hegde Rajesh 44
source and system level information fusion Mahanand
33 Joint Source Localization and Separation Using Sparsity Hegde Rajesh 45
Based Methods Mahanand
34 Preamble Based Channel Estimation in GFDM Banerjee Adrish 46
35 On Redundancy Rate of SWLZ and FDLZ Algorithms and Bansal Rakesh K 47

on Exponential Rates for Generalised AEP and Waiting
Times
36 Detection and Removal of Cracks from Digitized Artworks Gupta Sumana 48
37 Subspace Based Domain Adaptation for Visual Recognition Hegde Rajesh 49

Mahanand
38 Musical tempo Estimation using Sub-band Synchrony Hegde Rajesh 50
Mahanand
39 EMG modeling using Bessel series expansion Sircar Pradip 51
40 Regularity Flow Inspired Target Tracking in FLIR Imagery Venkatesh K S 52
41 Hybrid Structured Peer to Peer Live Video Streaming Singh YN 53

Protocol
42 Sparsity-Based Source localization under Sensor Position Hegde, Rajesh 54
Errors in Spherical Harmonic Domain Mahanand
43 Performance Enhancement with energy efficient Dynamic Singh Yatindra 55
Source Routing (DSR) protocol in Mobile Adhoc Network Nath
(MANET)
44 Anonymity in Delay Tolerant Networks Singh Yatindra 56
Nath
45 Fault Tolerant Clock Synchronization in Distributed Singh Yatindra 57
Network using Weighted Average Nath
4
46 Video Compression Using Video Decomposition Technique Gupta Sumana 58
47 Localization enhancement of wireless sensor networks Rajawat Ketan 59
48 Fast CFO estimation in OFDM systems using golden section Vasudevan 60

search algorithm Kasturi
49 Spatio-Temporal Spectrum Hole Detection with Multiple Rajawat Ketan 61
Primary Users
50 Preamble based Channel Estimation for an OFDM/OQAM Kasturi 62
system Vasudevan
51 EXIT Chart Based Performance Analysis Of Fixed Point Vasudevan 63
Implementation Of Turbo Codes Kasturi
52 Analytical Model for Information Propagation Speed in Singh Y N 64
Highway Vehicular Networks
53 QOS Analysis In Data Network: Stability, Reliability, QOS Singh YN 65
Invoke Rate Perspectives
54 Conversion of Aspect ratios using a novel mirror setup Venkatesh K S 66
55 Fault Tolerant Node Disjoint Multipath Routing In Manets SinghYN 67
56 PAPR Reduction In OFDM System By ZADOFF-Chu Vasudevan 68

Transform Kasturi
57 Anthropological Location & Identity Authentication Gupta Sumana 69
K S Venkatesh
58 Performance Analysis of a Novel Radar CFAR Detector Vasudevan 70
Kasturi
59 Performance of Channel Independent Precoder with and Vasudevan K 71
without Channel estimation for OFDM
60 Low complexity OFDM based on GIPS-Transform with Vasudevan 72
reduced BER and PAPR Kasturi
61 Algebraic Independence: Criteria and Structural Results Singh YN 73
over Diverse Fields
5
62 Interference Alignment With Channel Uncertainity And Sharma Govind 74
Estimated Channel Through Kalman Filter
63 Precoded Superimposed Pilot For Channel Estimation In Sharma Govind 75
Rayleigh And Rician Fading Channels: Uplink Massive
Mimo
64 Explorations Of Multidimensional Modulo TIC TAC TOE Venkatesh K S 76
65 Shape reconstruction for ISAR imaging Naik Naren 77
66 Marker Less Hand Detection And Gesture Recognition From K S Venkatesh 78

2d Sensor Data
67 Estiimation and Cancellation of Lens Flare Venkatesh K S 79
68 Precoding with time-domain feedback in MIMO-OFDM Vasudevan 80

systems Kasturi
69 On History Assisted Spectrum Sensing for Cognitive Radio Banerjee Adrish 81
70 On Techniques to Detect Malicious Users in Cooperative Banerjee Adrish 82

Spectrum Sensing
71 Image Binarization of Historical Degraded Document Gupta Sumana 83
Images Venkatesh K S
6
6
Microelectronics,VlSI &Technology Display

No No
1 Utilisation of heat generated in high concentration 2D Anand R S 84
photovoltaic systems by thermoelectric converters Iyer S SK
2 Thermal Resistance modeling in Fin FET and FDSOI Chauhan 85
Transistors Yogesh Singh
3 Simulatioin And Modelling Of Organic Light Emitting Diode Mazhari 86
Baquer
4 Study of suitability of four di-keto-pyrrolo-pyrrole derivatives Iyer S Sundar 87
for active layers in organic solar cell Kumar
5 Extraction and modeling of threshold voltage in organic thin Mazhari B 88
film transistors
6 Analysis of photo-response of top-contact OTFT Mazhari B 89
7 Study of suitability of four imidazolin-5-one derivatives for Iyer S Sundar 90

active layers in organic solar cells Kumar
8 Effect of annealing poly 3-hexylthiophene layers in the Iyer S Sundar 91
presence of pulsed electric field Kumar
9 Oscillator and Amplifier circuits based on single layer organic Mazhari B 92
devices
10 Compact Modelling Approach For Organic Thin Film Mazhari B 93
Transistors
11 Time Dependent Degradation Study of Performance of Mazhari B 94
P3HT:PC60BM and PTB7:PC70BM Based Organic Solar Cell
Devices
12 A Truly Passive Pixel Image Sensor Array Mazhari B 95
13 Effect of processing asymmetries on a-IGZO TFT Mazhar B 96
characteristics and OLED pixel driver circuits
14 Correlation between Dark & Light Characteristics of Bulk Mazhar B 97
Hetero-Junction Organic Photo-Voltaic Cells
7
15 A Study of Performance of Dual-Metal Gate Double-Gate Dutta Aloke 98
TFETs (DMG-DGTFETs) and Development of a new Surface
Potential based Drain Current Model
16 Development of a new Analytical Model for the Drain Current Dutta Aloke 99
of Double-Gate Tunnel FETs (DGTFETs)
17 Small-Signal Modeling of Laterally Asymmetric MOSFETs Dutta, Aloke 100
including Non-Quasi-Static Effects Chauhan Y S
18 Design and simulation of the thin film solar cell Copper Indium Iyer S Sundar 101
Gallium Selenide (CIGS) and Copper Zinc Tin Sulfide (CZTS) Kumar
using ATLAS , SILVACO
19 Electrothermal Numerical Modeling of heterogeneous binary Chauhan 102
transition metal oxide based RRAM Yogesh Singh
20 Modelling and Simulation of Laterally Asymmetric Channel Yogesh Singh 103
MOSFETS Chauhan
21 Investigation of Band to Band Tunnelling in Quantum-Well Iyer S Sundar 104
MOSFETs Kumar
8
RF Microwaves & Photonics
No No
1 Dual and Triple Band Bandpass Filter realization Using Split Biswas 105
Ring and Spiral Resonator for Significant Size Reduction Animesh
2 Studies on Multilayer Substrate Integrated Waveguide Slot Biswas 106

Coupler and its Application to Microwave Circuits Animesh
3 Studies on internal and external cloaks for improved cloaking Srivastava KV 107
performance using realizable material parameters
4 Design of X-Band Power Amplifier based on GAAS-PHEMT Biswas 108

Technology Animesh
5 Design of X-Band low phase-noise oscillator and comparative Biswas 109

analysis of the two topologies Animesh
6 Design of power amplifier Biswas A 110
7 Design of Broad Band Bias-T on Alumina Substrate Das Utpal 111
8 Design of broadband Bias-T on GAAS substrate Das Utpal 112
9 Design of Planar Resonator Sensor for RF Testing of Liquids Akhtar M J 113
10 Design of Multi Slot Vivaldi Antenna with Improved Akhtar M J 114

Directivity for Microwave Testing Applications
11 Continuously Tunable Multi-wavelength Ultra-Fast Fiber Ring Kumar Pradeep 115

Laser
9
12 An Inverse Scattering Approach To Design Small Size 116
Microstrip Filters Using Defected Ground Structures Akhtar M J
13 Fractal Geometry Inspired Polarization Insensitive Microwave Srivastava 117

Absorbers for Multiband and Broadband Applications KV
14 Analysis and Design of Ultra Wideband Wearable Antennas Harish A R 118
15 Compact And Enhanced Bandwidth Polarization Harish A R 119

Reconfigurable E-Shaped Patch Antenna
16 Design, Fabrication and Characterization of a Vacuum Gated Das Utpal 120

Diode
10
Power &Control

No No
1 Design, implementation and control of low voltage DC Das Shyama 121
microgrid Prasad
2 Hall Effect Current Sensor using Transconductance Sensarma 122
Amplifier PSarathi
3 Observability analysis and topology error processing of Chakrabarti 123
power systems in the presence of hybrid measurements Saikat
4 Optimal demand response for PV-Integrated households Chakrabarti 124
Saikat
5 Online Health Monitoring of Aluminium Electrolytic Anand 125
Capacitors For Single Phase Grid Connected PV System Sandeep
6 An Optimal Smoother Based Dynamic State Estimator for Chakrabarti 126
Power Systems Considering PMU Measurements Saikat
7 Tunnel FETs and Its Application to Digital Circuits Mishra 127
Santanu
8 Power Sharing in Low Voltage DC Microgrid Using Low Anand 128
Speed Communication Sandeep
9 Zero Net Energy House Cluster: Protection and Energy Singh S N 129
Management
10 Development of Control Strategies for Power Management Srivastava 130
with Renewable Resources SC
11 Object Matching Based on Speeded Up Robust Features Verma 131
NK
12 Rotor Angle Estimation and Coherency Detection of Chakrabarti 132

Synchronous Generators Saikat
13 Sliding Mode Control Strategies for Robotic Systems Behera 133
Laxmidhar
14 Vision Based Navigation of a Quadcopter using a single Behera 134
camera Laxmidhar
11
15 Voltage Stability monitoring using Lyapunov Exponent Chakrabarti 135
Saikat
16 Power Quality Improvement and Power Capacity Mishra 136
Enhancement of Power Plant for Rural Telecom Santanu
Applications Joshi Avinash
17 Modeling, Analysis, and Implementation of a Hybrid MOC Mishra 137
for Solar PV Applications Santanu
18 Design and analysis of ultra-lift LUO converter based PFC Mishra 138
circuits Santanu
19 Series Active Filter for Rectifier DC Link Current Sensarma 139
Partha Sarathi
20 Design and Development of Single Phase Dynamic Voltage Anand 140
Restorer Sandeep
21 Study of Silicone Rubber based Composites with Nano- Gupta 141

sized Fillers Nandini
22 Design, Simulation and Development of a Multifunctional Das S P 142
Active Power Filter for Power Distribution System
23 Design and Modelling of VSC-HVDC Transmission System Singh S N 143
Having Wind Farm
24 DSP based Implementation of Induction Motor Control with Das S P 144
Estimation of Rotor Time Constant
12
Title : Video stabilization and object tracking using AFFINE-
Spatio temporal regularity flow
Author(s) : MD Afaque Azam
Roll No : 12104041
Supervisor(s) : Gupta Sumana
Abstract:
Today, hand held devices are present everywhere. Every mobile phone today comes with an
inbuilt camera. This has led to a sharp increase in the amount of videos created by
nonprofessional users. The jitters caused by this unintentional motion of the video cameras is
very common . Such videos are not only used by common users but also by scientists. Launch
pads have a large number of cameras that record footage of space vehicles and the surrounding
areas during launches. While they take off, a very large vibration is produced due to thrust. This
causes nearby cameras to shake. Aside from making the footage difficult to watch, the
instabilities in the background make it hard for viewers to notice any debris that may have fallen
from the vehicle, or any other foreign objects. Because of these and many other reasons, video
enhancement has become very important. A particular requirement is to stabilize the video. The
process of electronically removing this unwanted jitters from the video is called video
stabilization. A new model called Spatio-Temporal Regularity Feature(SPREF) is introduced.
This feature finds the direction in which sum of directional gradients of pixel is minimum. In
other words it gives the direction along which intensity variation of pixel is minimum. Translated
box splines are used to approximate the flow curves. We model the video using Affine SPREF
and get the regularity direction of each pixel. Using this model we find the global motion of the
frames and hence the motion of camera. We perform Kalman filtering to remove unwanted jerks
from the intentional motion. Finally we perform image warping to get back stabilized frames in
the final stage of video stabilization. We present our result on several test videos. Also a
comparison of some standard methods is done using ITF parameter which is explained in the
same section. We extend the application of Affine SPREF to object tracking. Object tracking
finds several real world applications like Automated video surveillance, Robot vision, Traffic
monitoring, Animation etc. We perform foreground segmentation by thresholding the SPREF
energy of voxels, of the octree segmented video cubes. The resulting segmented videos are used
to keep track of desired moving objects.
For more details click here back
13
Title : Modeling Twitter Hashtag Trends
Author(s) : BANSAL SAURABH
Roll No : Y9227524
Supervisor(s) : Jagannatham Aditya K
Abstract:
Microblog (e.g. Twitter) data is an established social medium for information di usion. Patterns
of information di usion through such platforms are limited. This thesis presents a method to
model such patterns of information di usion using temporal aspects only. We have used the
Infochimps dataset which contains the time series for a 5% sample of all tweets between Jan
2008 to Dec 2009. Our analysis unveils that a hash tag's popularity is driven by a combination of
lagged time series of the hash tag and the general behavior of the micro blogsphere. Based on
this observation we analyze the contribution of di erent variables on popularity and describe a
predictive modeling technique for the same using forward step by step ordinary least squares
regression. Our analysis shows that the popularity is majorly driven by lagged time series(
explains 60% of the variance).
14
Title : A Distributed Message Dissemination Framework for
Emergency Response Systems over Smartphones
Author(s) : Bhatnagar Abhiroop
Roll No : 10327015
Supervisor(s) : Sharma Govind
Abstract:
In this thesis, we propose a framework for distributed dissemination of messages by leveraging

the capabilities of smartphones. The major contribution of this work is an implementation of a
broker-based peer to peer protocol for prioritized delivery of messages in order to support a fast
emergency alert and response mechanism. The fundamental challenge lies in adapting the
community inspired publish subscribe model in privacy preserving and scalable manner for high
mobility scenarios. Our approach is based on mechanism of dissemination of messages through
streaming channels such as Google Cloud Messaging. As social media generated interactions are
uncontrolled, they cannot connect directly with an emergency alert and response mechanism. So,
we have implemented a distributed device information repository for controlled dissemination of
messages along with a framework for handling responses appropriate to alert messages. The
two key features we tried to preserve in implementation of our protocol are (i) energy awareness,
and (ii) prioritized delivery of messages. In particular, in this thesis report our focus is on the
origin and theoretical rationale of system design and the implementation of the Data Cart
Tracker.
15
Title : Learning Distributed Document Representations for Multi-
Label Document Categorization
Author(s) : Gupta Nitish
Roll No : 10327461
Supervisor(s) : Hegde Rajesh Mahanand
Abstract:
Multi-label Document Categorization, the task of automatically assigning a text document into
one or more categories has various real-world applications such as categorizing news articles,
tagging Web pages, maintaining medical patient records and organizing digital libraries among
many others. Statistical Machine Learning approaches to document categorization have focused
on multi-label learning algorithms such as Support Vector Machines, k-Nearest Neighbors,
Logistic Regression, Neural Networks, Naive Bayes, Generative Probabilistic Models etc. while
the input to such algorithms i.e. the vector representation for documents has traditionally been
used as the bag-of-words model. Though the usage of simple bag-of-words representation gives
surprisingly accurate results, it suffers from sparsity, high-dimensionality, lack of similarity
measures along with other drawbacks such as the inability to encode word ordering and
contextual information in which the words occur. Encoding contextual information about words
in documents is crucial to capture the correct semantic content of the highly complex and
ambiguous human language. Our work is focused on learning continuous distributed vector
representations for documents by embedding all the documents in the same low-dimensional
space such that documents that are similar in their semantic content have similar vector
representations. To tackle the issues in bag-of-words representation model, we present an
unsupervised neural network model that uses the document vector to predict words in the
document along with using the contextual information in which the word occurs and jointly
learns distributed document and word representations. We develop a modified version of the
logistic regression algorithm to learn similar distributed representations for categories to perform
the document categorization task. We show that our model gives state-of-the-art results on the
standard Reuters-21578 dataset, improving the bag-of-words model by 9% and previous state-of-
the-art by 3.26% in terms of the F1 Score. We also show the effectiveness of our model in
imputing missing categories on the Wikipedia articles against the bag-of-words representations.
As we embed documents, categories and words in the same low-dimensional space our model
can also estimate semantic similarities between them. We qualitatively demonstrate that the
learned representations capture the semantic dependencies between categories and words which
is not directly observed in the data.
16
Title : HDR imaging with conventional cameras
Author(s) : Nawhal Megha
Roll No : 10327397
Supervisor(s) : Venkatesh K S
Abstract:
Real world scenes have huge intensity variations which are not in the control of the capture
process. While the human eye has an excellent dynamic range that enables us to visualize precise
contrast variations, and dynamically adapts to illumination variations, the dynamic range of the
conventional imaging devices is limited because of the physical constraints of the sensors. Image
saturation is observed often when lighting conditions are very bright. In such scenarios, the
captured image will have some optimally illuminated parts while some parts may undergo
saturation (underexposure or overexposure). This is due to the gray level clipping of the camera
sensor. The consequence of this clipping is complete loss of detail in the saturated regions, which
makes the image aesthetically unappealing for human use as well as unsuitable for machine
consumption, because of the disappearance of all feature information. In this work, we propose a
solution to recover the scene information lost due to saturation, and hence produce a better
quality image. This final output high dynamic range(HDR) image is obtained by blending a
sequence of gain-shifted low dynamic range (LDR) images. We also put forth an idea of solving
the problem of image saturation during the acquisition process at pixel level to achieve resultant
image with no or minimal saturation.
17
Title : People counting in high density crowds from still images
Author(s) : Bansal Ankan
Roll No : 10327095
Abstract:
In this work, we present a method for estimating the number of people in high density crowds
from still images. The method estimates counts by fusing information from multiple sources.
Most of the existing work deals with very small crowds (tens of individuals) and uses temporal
information from videos to estimate the counts. Our method uses only still images to estimate the
counts in high density images (hundreds to thousands of individuals). At this scale, several
problems including occlusion, perspective, few pixels per person and clutter render counting by
human detection infeasible. Also, we cannot rely on only one set of features for count estimation.
We, therefore, use multiple sources, viz. interest points (SIFT), Fourier analysis, wavelet
decomposition, GLCM features and low confidence head detections, to estimate the counts. Each
of these sources gives a separate estimate of the count along with confidences and other
statistical measures which are then combined to obtain the final estimate. We tested our method
on an existing dataset of fifty images containing over 64000 individuals. Further, we added
another fifty annotated images of crowds and tested on the extended dataset of hundred images
containing over 87000 individuals. The counts per image range from 81 to 4633. We report the
performance in terms of mean absolute error, which is a measure of accuracy of the method, and
mean normalised absolute error, which is a measure of the robustness.
18
Title : Flash crowd handling in P2P live video streaming systems
Author(s) : Dwivedi Anurag
Roll No : 10327132
Supervisor(s) : Singh Yatindra Nath
Abstract:
Peer-to-peer systems have greatly enhanced live streaming experience by creating efficient and
highly scalable streaming overlays where bandwidth capabilities of all peers can be utilized.
However, realization of such systems have been challenged by the phenomenon of flash crowd -
the arrival of hundreds of thousands of peers in a very short span of time. Such situations may
typically arise at the beginning of live streaming events such as a football match or a live lecture.
Experiments have shown that the system can scale only up to a limit during flash crowd. This is
limited both by the available surplus bandwidth as well as the intense competition among the
peers for scarce initial resources. Various population control measures have been suggested for
both mesh-based and tree-based live streaming systems. The main focus area of this thesis is on
tree-based systems. Such systems have seen some centralized solutions. In this thesis, a
distributed algorithm with minimal central control is presented which organizes the newly
arrived peers into hierarchical positions to reduce competition among them. This hierarchical
rank is then used to construct different sub-stream trees. The video stream is divided into sub-
streams and each sub-stream is pushed over a separate sub-stream tree. Only the peers at the top
of the hierarchy will directly access the scares initial resources and in turn forward the stream to
those below them in the hierarchy. Thus, by utilizing the resources provided by the newly arrived
peers in handling ash crowd, better system scale can be achieved.
19
Title : SBL based joint sparse channel estimation and maximum
likelihood sequence detection in OSTBC MIMO-OFDM
systems
Author(s) : N S Yashaswini
Roll No : 13104161
Abstract:
In this work, we begin with sparse representation of the wireless multipath channel in an
orthogonal space-time block coded (OSTBC) multiple-input multiple-output (MIMO) or-
thogonal frequency division multiplexing (OFDM) system owing to the fact that the number of
active components in the multipath power delay channel profile are very less in comparison to
the overall delay spread. Based on this, we additionally consider a challenging ill-posed channel
estimation scenario in which we exploit the inherent temporal sparsity and employ the sparse
Bayesian learning (SBL) framework to develop a robust pilot-based channel estimation
scheme. Furthermore, we also propose a novel data aided SBL-based joint sparse channel
estimation and maximum likelihood sequence detection (MLSD) technique based on the
expectation maximization (EM) framework and illustrate that the seemingly intractable joint
estimation paradigm finally reduces to an SBL-based sparse channel estimate in the E-step
followed by a novel modified ML decision metric-based sequence estimate in the M- step. We
also present a comprehensive Bayesian Cramer-Rao bound (BCRB) analysis of the proposed
SBL-based schemes and present a detailed complexity analysis of the proposed scheme. Finally,
the simulation results are presented to validate the theoretical bounds and illustrate the
superiority of the proposed techniques.
20
Title :
A novel time domain adaptive decision feedback qualization
technique for cyclic code shift keying modulated wireless
systems and convergence analysis of LMS and RLS
algorithms
Author(s) : Vajjaramatti, Harshavardhan
Roll No : 13104051
Abstract:
Cyclic Code Shift Keying(CCSK) is form of spread spectrum technique mainly used in
applications such as military, global positioning systems(GPS), wireless local area network(
WLAN) etc. Due to the frequency selective nature of the channel there is a problem of inter
symbol interference, and if the channel is also time varying due to relative motion between
transmitter and receiver, then it will be difficult to track channel variations, this motivates us to
design adaptive channel equalization to solve above problems. In this work, we begin with the
design of minimum mean square error(MMSE) and zero forcing(ZF) linear and decision
feedback equalizers in time domain for CCSK modulated wireless systems assuming the perfect
channel state information(CSI) at the receiver. Further the linear and decision feedback
equalizers are developed based on the optimal delay which enhanced the mean square
error(MSE) performance.In the second part of the thesis work, a novel time domain adaptive
decision feedback equalization(ADFE) and adaptive linear equalization(ALE) receiver structures
has been developed by employing least mean square(LMS) and recursive least squares(RLS)
algorithms without CSI. The convergence analysis of LMS and RLS algorithms have been
carried out based on the analysis of ensemble average learning curve for different step size
parameters corresponding to LMS and RLS algorithms. Next unlike conventional analysis for
unity forgetting factor, in this work the steady state error and misadjustment in RLS algorithm
have been addressed for forgetting factor less than unity.Extensive simulations are carried out to
demonstrate the MSE performance of LMS and RLS algorithms corresponding to different step
size parameter and forgetting factor under moderately and severely distorted channel conditions.
The theoretical and simulated curves are seen to be in close agreement with each other. Finally
the simulations for bit error rate(BER) performances for both receiver architectures are carried
out, and it is observed that for severely distorted channel conditions the novel decision feedback
equalizers demonstrates the enhanced MSE and BER performances in comparison to that of the
linear equalizers.
21
Title : Multiple beacon based robust decentralized detection in
cooperative mimo wireless sensor networks
Author(s) : Ram Hukma
Roll No : 13104054
Abstract:
In this thesis work, we consider the decentralized detection in cooperative multiple-

input multiple-output (MIMO) wireless sensor network (WSN) which consists of multiple sensors
and a fusion center. All the sensors and the fusion center are equipped with the multiple transmit
and receive antennas respectively. We formulate the multiple beacon based novel decision fusion
rules for two scenarios. In the rst scenario, the channel aware decision fusion rules has been
derived where the fusion center is having perfect channel-state-information (CSI). The channel
aware fusion rule is derived for two signalling schemes employed between the sensors and the
fusion center namely antipodal signalling and non-antipodal signalling. In this scenario there is a
stringent requirement of perfect CSI. Due to dynamic nature of wireless media one can only have
CSI with limited accuracy. Accordingly, in the second scenario, the generalised likelihood ratio
test (GLRT) based robust decision fusion rule has been derived which incorporates the
knowledge of uncertainty present in MIMO channel estimates at the fusion center. The GLRT
based robust fusion rule is derived for antipodal signalling employed between the sensors and the
fusion center. Obtaining closed form expression for the GLRT based robust fusion rule with non-
antipodal signalling is mathematically challenging and which has to be worked out in future. The
analytical expressions for detection performance analysis parameters, i.e. probability of detection
and probability of false alarm, at the fusion center has been derived for channel aware fusion
rules. For the purpose of detection performance comparison receiver operation characteristics
(ROCs) have been plotted considering di erent simulation parameters. Simulation results show
that the channel aware detector performs best among all the proposed detectors and the GLRT
based robust detector yields superior detection performance compared to the uncertainty agnostic
detector.
22
Title : Interference Alignment For Linear Coherent Decentralized
Parameter Estimation In Multiple Wireless Sensor Networks
Author(s) : Gorain Kanchan Kumar
Roll No : 13104063
Abstract:
In this work, we have considered linear distributed estimation of Gaussian random vector
parameter when multiple sensor networks are interfering each other. To nullify the inter network
interference and efficiently use the available channel bandwidth with maximum throughput,
interference alignment scheme is used. We have derived the BCRB (Bayesian Cramer Rao
Bound) for the estimation of the vector parameter at the fusion center and it has been shown that
LMMSE (Linear Minimum Mean Square Error) estimate achieves the BCRB for considered
scenario when perfect interference alignment is achieved. We have also considered the total
power constraint of a sensor network and have presented a close form expression for precoded
amplify and transmit, precoded equal and precoded optimal power allocation scheme with
interference alignment when the sensor observations are noiseless. However for noisy sensor
observation the optimal power allocation is left as a future work. We further propose that sensor
selection can significantly improve the performance of the system when considered in
interference alignment framework and simulation result verify the claim. For sensor selection we
have further proposed optimal exhaustive search sensor selection algorithm and two low
complexity algorithm named as threshold based sensor selection and greedy sensor selection
algorithm. The simulation results are also carried out to compare the performance and
complexity of the algorithms.
23
Title : Battery Management Through Real Time Pricing
Author(s) : Pullela Venkata Aditya Prasad
Roll No : 13104107
Supervisor(s) : Rajawat Ketan
Abstract:
The introduction of smart devices, controllable batteries, renewable power sources, and smart
metering systems allows users the opportunity to optimize their energy consumption on a day-to-
day basis. Towards this end, utility companies are planning to introduce real-time pricing (RTP),
wherein price per unit of electricity will be time-varying and depend on the average load profile
and availability of resources. In this work, we consider the problem of storage management at
the household level with RTP. Existing works formulate a complicated convex optimization
problem to solve a simpler day-ahead pricing problem. The markov decision framework allows
unknown prices, but is still too complicated to implement on a smart meter. We propose a real-
time battery management system that outputs charging and discharging decisions per time slot.
The algorithm utilizes the estimated state-of-charge (SOC) for taking decisions, and has an order
of magnitude lower complexity than existing schemes. The mathematical model for the
algorithm is developed, and the steady-state behavior is studied. Practical considerations, such as
adaptive SOC estimation, on line load scheduling, excess loading, and renewable integration are
also incorporated within the framework. Extensive simulations show the superior performance of
the proposed algorithm over existing ones
24
Title : Spatiotemporal Colorization of Video Using 3D Steerable
Pyramids
Author(s) : Paul Somdyuti
Roll No : 13104143
Abstract:
Colorization refers to the process of producing a coloured image/video from a given

monochrome image/video. In this work, we develop a new approach for video colorization
through spatiotemporal propagation of colour from the colour scribbles which are marked by the
user on a selected set of keyframes. The keyframe selection is performed by adaptively
thresholding a dissimilarity measure derived from the change in dominant orientation response of
steerable filters with respect to the nearest keyframe. The spatiotemporal colour propagation
within the video volume is carried out using the dominant orientation responses of the output of a
set of 3D steerable filters, steered to a predetermined number of orientations. The 3D steerable
filters are used to perform a steerable pyramid decomposition of the video, and the dominant
orientation response at each pixel of the video are found at each level of the pyramid. Since
information about motion as well as texture of the video are embedded in the steerable pyramid,
our approach eliminates the need of motion vectors, which are normally used to transfer colours
from one frame to the next in the conventional video colorization techniques. This results in
faster computation, and greater accuracy of colorization. At each level of the pyramid, priorities
are assigned to the pixels based on the presence of other coloured pixels in the vicinity, as well
as the spatiotemporal smoothness of the neighbourhood. Pixels are colorized in the decreasing
order of priorities starting with the topmost level of the pyramid. At the topmost level of the
pyramid 3D patches belonging to the spatiotemporally smoothest regions of the video are
colorized, which correspond to static or almost static regions of the video, with relatively low
texture content, followed by regions with more motion and texture content at the lower levels of
the pyramid. Experimental results demonstrate that our approach can effectively colorize videos
with different types of motion and visual content with better accuracy and lesser computational
requirements as compared to the conventional video colorization techniques.
25
Title : Stochastic resource allocation in device to device
communication
Author(s) : Kandula Surya Viswanadh
Roll No : 13104065
Abstract:
With the rapid proliferation of internet-connected mobile devices, the demand for high data rates
has increased beyond what base stations (BS) can handle. A new technology in this regard is
Device to Device (D2D) communications, where the BSs can offload its traffic to pairs of
neighboring mobile users. An important problem in this context is that of mode selection, where
one must decide if the data is to be routed through the BS or directly. In either case, power and
rate allocations must also be decided, based on the available channel state. This thesis formulates
the joint resource allocation and mode selection problem in D2D communications. Assuming full
channel knowledge at the BS, the problem is posed as that of maximizing the utility, subject to
stochastic rate constraints. We solve this optimization problem using Stochastic subgradient
algorithm in dual domain. On an average, the allocated resources converge to their near optimal
values if the step size parameter is close to zero. The simulation results proved the efficacy of the
proposed algorithm.
26
Title : Calibration and mapping for soot monitoring
Author(s) : Singh Pradyumn
Roll No : 13104100
Abstract:
Soot is a major pollutant emitted from manufacturing industries, biomass burning and wood &
crop residues due to incomplete burning of hydrocarbons. Black carbon (BC) and Elemental
carbon (EC) are major component of soot and a prime contributor to global warming. To asses
and control it's regional impact in remote locations, comprehensive and low-priced measurement
techniques are needed. A developed photo-reference method of measuring soot remotely is
economical, wireless and requires low-power. In this theses, we have developed optimal
regression model to describe the relationship between surface Black Carbon of a an aerosol
loaded quartz filter and reflectance value (pixel value)of captured image of same loaded filter.
We have made use of an special filter holder instrument which is used to keep the loaded filter in
some special light conditions and a digital camera is used to take the picture of the loaded filter.
Regression analysis is employed to choose the best regression model which can be used for
future predictions of BC concentration most accurately using image pixel values of loaded filter
image (linear or non-linear) among candidate models. In particular, Akaike Information Criterion
selects the regression model which minimizes the statistical distance between true model and
chosen model. We showed Red & Green pixel value of loaded filter image in presence of red
light inside the filter holder instrument are among best predictors as compare to others. In the
later part of this theses, the problem of mapping of pollution concentrations amid known sampler
locations, using known concentrations at sampler locations is considered. A three layer Artificial
Neural Networks called Radial Basis Function Network as Gaussian activation functions is
brought in play for smooth interpolation of pollution amid sampler locations. It is shown that the
prediction error is acceptable even without full knowledge of the environmental parameters.
27
Title : Channel Tracking and Estimation Transmit Beamforming
With Frugal Feedback using KF And Quantized-KF
Approach
Author(s) : Kumar Rajeev
Roll No : 13104113
Abstract:
Channel state information at the transmitter (CSIT) is important for a large number of
communication systems that rely on transmit beamforming or interference alignment for
increasing the signal-to-interference-plus-noise ratio. The use of CSIT however incurs a huge
amount of overhead, since CSI can only be estimated at the receiver, and must be fed back to the
transmitter. It is possible to reduce the amount of feedback severaly in slow-fading
environments, where the channel variations are not fast enough. Towards this end, a sign- of-
innovations Kalman lter (SOI-KF) has been proposed that allows the receiver to simply feedback
the sign of the innovation term appearing in the Kalman lter at each time slot. Thus, the SOI-KF
incurs an overhead of one bit per block. A number of quantized KF (Q- KF) have also been
proposed, where 2 or more bits can provide better performance. This thesis proposes a quantized
version of KF (Q-KF) that aims to reduce the feedback overhead to less than one bit per time
slot. Towards this end, the idea is to send no feedback when the innovation term falls below a
threshold. The proposed QKF provides a continuous trade o between the communication cost
with performance. Analytical results for the MSE of the proposed algorithm are also presented
and compared with the simulation results.
28
Title : Distributed cooperative localization via ADMM
Author(s) : Jain Rahul
Roll No : 13104110
Abstract:
Localization is the problem of primary importance in wireless sensor and ad-hoc networks.
Accurate and low cost localization of nodes is important in many applications, as in most of the
cases the data collected from the sensor is meaningful only with the relevant location.
Mathematically however, the cooperative localization problem is non-convex, and therefore
difficult to solve in a distributed fashion. This thesis introduces an ADMM-based algorithm for
solving non-convex problems in a distributed scenario. The proposed algorithm is partially
asynchronous, and can handle some erroneous or delayed updates. Using mild assumptions,
convergence to the stationary point is shown. We have applied the algorithm to the cooperative
localization problem where the nodes converge to locally optimum coordinates via consensus.
Computer simulations show that the performance of the proposed algorithm is better then the
existing synchronous as well as asynchronous algorithms.
29
Title : GLRT based target detection in monostatic MIMO RADAR
systems
Author(s) : S Jyothsna
Roll No : 13104061
Abstract:
This work presents a target detection technique in mono static MIMO radar when the RCS
coefficients of the scattering scene is deterministic and unknown quantity. GLRT based test
statistics is derived in which scattering scene coefficients are estimated by using maximum
likelihood (ML) estimation. Further the above framework is also extended for a scenario with
unknown noise variance. Also, the asymptotic detection performance for the above detector is
obtained. The ML estimation used to estimate the scattering scene RCS coefficients, achieves the
CRLB bound. We have also extended the GLRT based detection statistics for a framework with
reduced number of range cells , to increase the angular resolution at the cost of slight increase in
complexity. The performance of the GLRT detector is compared with that of energy detector and
the proposed detector shows a better performance in terms of ROC. The above detection
paradigm is also extended for moving target detection scenario. We have employed an iterative
minimization of least squares error for the case where the entire scattering scene under
consideration moves with the same velocity to find the radial velocity of target. Simulations are
presented to validate the results.
30
Title : A Novel Bandwidth Efficient Channel Estimation Scheme
In Ostbc-Mimo-Ofdm Systems Using Superimposed Pilots
Author(s) : S Sagar
Roll no : 13104129
Abstract:
This work proposes a novel bandwidth efficient channel estimation scheme which utilizes only a
fewer number of pilot sub-carriers by employing the superimposed pilots (SIP) based strategy. A
novel precoding and decoupling matrices are designed to nullify the unknown data and the SIP
during channel estimation and data detection, respectively. The proposed scheme is employed on
spatially uncorrelated channel to perform channel estimation in time domain using minimum
mean square error (MMSE) and least square (LS) estimators. For each of these estimators, a
novel tractable closed form optimal SIP structure are designed which minimizes the mean square
error (MSE) under the constraint of total transmitted SIP power. Cramér-Rao bound (CRB) and
Bayesian Cramér-Rao bound (BCRB) are derived for LS and MMSE estimators, respectively. A
closed form exact bit-error-probability (BEP) expression is derived for the square orthogonal
space time block coded multiple-input-multiple-output orthogonal frequency division
multiplexing (OSTBC MIMO-OFDM) systems in the presence of channel estimation error.
Comparing to the existing scheme, simulations demonstrate that, for an MMSE estimator, the
proposed scheme performs similar in terms of MSE and bit-error-rate (BER) under the constraint
of equal total pilot power and equal data power per symbol and outperforms in terms of
bandwidth (BW) efficiency.
31
Title : Optimal Target Detection and 3-D MUSIC based Parameter
Estimation for Bistatic MIMO Radar Systems
Author(s) : Mittal Neha
Roll No : 13104090
Abstract:
This work presents novel schemes for multi-dimensional target detection and estimation for a
bistatic multiple-input multiple-output (MIMO) radar system with uniform linear array at the
transmitter and the receiver. We have developed an optimal detector in Neyman Pearson sense
for deterministic unknown radar cross section (RCS) and have derived the closed form
expressions for probability of detection and probability of false alarm. Maximum Likelihood
estimation has been used to estimate the RCS. It is extended to the case when target RCS is a
random quantity. Receiver operating characteristics (ROCs) have been plotted for different
values of number of transmit/receive antennas and is compared with energy detector. Further, we
have discussed the joint estimation problem of direction of departure (DOD), direction of arrival
(DOA) and Doppler frequency using multiple signal classification (MUSIC) algorithm. We have
proposed a reduced dimension algorithm for estimation of the above mentioned parameters for
multiple targets. The advantage of proposed scheme is that it provides automatic pairing. Finally,
simulations are presented to validate the results.
32
Title : Robust precoder design for decentralized estimation in
wireless sensor networks
Author(s) : Narayana B Bhargav
Roll No : 13104025
Abstract:
In this work, we present a robust precoding scheme towards linear decentralized estimation of
source parameter in a wireless sensor network (WSN). We consider a coherent multiple access
channel between the sensors and the fusion center. The actual channel state information between
the sensor nodes and the fusion center are imprecise and it is modeled to be located in an
ellipsoid centered about the channel estimate. First we consider the problem of finding the robust
precoder when the source parameter to be estimated is a scalar. We propose a worst-case design
approach where the mean square error of the parameter is minimized for the least favorable
channel located in an ellipsoid centered about the estimated channel in a Single Input Single
Output (SISO) scenario. Next, we design robust precoders for Multiple Input Multiple Output
(MIMO) WSN to estimate a vector parameter by assuming that the observations at the sensor
nodes are noiseless. The proposed schemes result in an optimal parameter estimate at the fusion
center with minimal processing at the fusion center. The robust precoder design is cast as convex
optimization problems with second-order conic constraints which can be solved efficiently with
existing tools.
33
Title : Analysis Of Spatially Modulated Systems For Spectrum
Sensing And Cooperative Communication Applications
Author(s) : Goel Amish
Roll No : 10327071
Abstract:
In this work, we analyze the performance of spatially modulated systems for the problems of
spectrum sensing and cooperative transmission. Spatial Modulation is a recently proposed
technique that aims to strike a trade-off between energy efficiency and spectral efficiency of the
communication system. Several novel schemes are proposed for spectrum sensing of the
cognitive radio systems with spatially modulated (SM) multiple-input multiple-output (MIMO)
primary users. A composite hypothesis testing approach is used to model the unknown
parameters in the system. In particular, the generalized likelihood ratio test (GLRT) based
detector is initially developed, followed by the correlator, low complexity Linear GLRT and the
energy detector (ED). Further, closed form expressions are derived for the probability of false
alarm and probability of detection to characterize the spectrum sensing performance of the
proposed schemes, employing also some results from extreme value theory. Simulation results
are presented to demonstrate the performance of the proposed spectrum sensing schemes for SM
systems and verify the analytical results. Next, the MIMO cooperative systems, with one relay
node and using selective decode and forward (DF) protocol for transmission at relay, are
investigated. In particular, the performance of the system is analyzed when the source and relay
employs space-time block coded spatial modulation (STBC-SM) scheme for transmission. The
system performance is compared with cooperative Alamouti-STBC scheme and the non-
cooperative transmission of the codewords. The high signal to noise ratio (SNR) performance is
investigated to derive the diversity order of the system. Finally, the optimal power allocation
between the source and relay is derived and simulation results are provided to show the
improvement over the uniform power allocation scheme
34
Title : Performance analysis of decode and forward protocol based
cooperative communication systems
Author(s) : Dadige Sree Moorthy
Roll No : 13104035
Abstract:
In this thesis, we study the performance of selective decode-and-forward relaying based MIMO
cooperative communication system by employing the quasi orthogonal space time block
codes(QOSTBC) at both source and relay nodes. We also show that performance of the system is
improved by employing QOSTBC coding when compared to orthogonal space time block
codes(OSTBC) for more than two transmit antennas. A comprehensive analysis of the end to end
pairwise error probability is provided for Jafarkhani, TBH and SP kind of the QOSTBC coding
schemes. We then derive the asymptotic approximation of the PEP which gives the insight of the
diversity order of the system. In addition, we investigate the selective DF relaying based free
space optical(FSO) cooperative communication system over the long range(weak,moderate and
strong) of atmospheric turbulence conditions. We have considered the Gamma-Gamma
distributed Irradiance factor for the purpose of analysis and to account all the possible
atmospheric turbulence conditions. We have derived an upper bound on the SER of the system
for MPSK modulation scheme and calculated a very close approximate SER expression for
BPSK modulation scheme. Finally concluded the work by deriving an expression for diversity
order for the above system.
35
Title : Multi-User Beamforming in Frequency-Selective MIMO
MC-CDMA systems
Author(s) : Dadhania Vishwa Chandulal
Roll No : 13104034
Abstract:
In this thesis, we intend to deal with the multi-user interference (MUI) present in
frequencyselective multiple-input multiple-output multi-carrier code division multiple access
(MIMO MC-CDMA) wireless systems. Design of precoders in frequency-selective scenario will
yield an intricate structure which is difficult to solve. Unlike the computationally inefficient and
conventional approach of considering the equivalent multi-user multiple-input multiple-output
(MU MIMO) channel in the frequency domain, we attempt to design novel precoders for
elimination of the MUI in time domain. In this direction, the multi-path multi-carrier decorrelator
(MMD) receiver structure is employed at each user to exploit the available multi-path diversity
of the frequency-selective channel which successfully reduces the MIMO MC-CDMA system
under consideration into an equivalent frequency flat MU MIMO system with appropriate
dimensions. Thereby, it opens up a way to apply various transmit preprocessing techniques
which can efficiently cancel the MUI from the received data at each user. Firstly, a well-known
generalized form of channel inversion namely the “block diagonalization(BD)” technique,
suitable for multi antenna receivers is studied. It is observed that while on one hand MMD
receiver scales up the dimensions in terms of number of available multipath components, the BD
technique on the other hand strictly works for nulling complete MUI in the system as a
consequence of which the number of users that the system can support reduces. To overcome this
shortcoming, we develop a joint path and antenna selection algorithm which performs selection
of receive antennas from the set of all available multipaths at each user and feeds back the
selected paths to the base station(BS) for the design of the appropriate zero-forcing precoders.
Secondly, the successive optimization framework is considered wherein the “successive
constrained eigenbeamforming (SCEB)” technique exerts a comparatively less stricter constraint
on the number of active users in the system since it does not concentrate on removing the
complete MUI present in the system. In this context, the study of the various preprocessing
techniques used in conjunction with orthogonal codes finally leads us to provide some useful
insight on the sum rate capacity achieved and the number of users successfully supported in a
MIMO MC-CDMA wireless system.
36
Title : Depth estimation from single image using texture and
defocus cues
Author(s) : Srikakulapu Vivek
Roll No : 13104146
Supervisor(s) : Gupta Sumana&Venkatesh K S
Abstract:
The challenging problem of depth estimation from single 2D image is an active field of research
in image processing. The motivation behind this work is the application of depth estimation for
3D scene reconstruction. As imaging is a process of 2D projection of a 3D scene, the depth
information is lost at the time of image capture from conventional camera. This depth
information can be inferred back from a set of visual cues present in the image, namely
monocular and binocular depth cues respectively. In this thesis, we present a model that
combines two monocular depth cues namely Texture and Defocus, for converting the 2D image
back to 3D image. Depth is related to the spatial extent of the defocus blur by assuming that
more an object is blurred, the further it is from the camera. At first, we estimate the amount of
defocus blur present at edge pixels of an image. This is referred to as the Sparse Defocus map.
Using the sparse defocus map we generate the full defocus map. However such defocus maps
always contain hole regions and ambiguity in depth. To handle this problem an additional depth
cue, in our case texture has been integrated to generate better defocus map. Texture integration in
defocus map estimation mainly focuses on modifying the erroneous regions in defocus map by
using the texture energy present at that region. The sparse defocus map is corrected using texture
based rules. The hole regions, where there are no significant edges and texture are detected and
corrected in sparse defocus map. We have used region wise propagation for better defocus map
generation. The accuracy of full defocus map is increased with the region wise propagation. The
main contribution of this work lies in achieving an increased accuracy of defocus estimation by
hole filling, integrating texture information with defocus cue and using region wise propagation.
37
Title :
Reversible Watermarking in Images using Difference
Expansion
Author(s) : Patel Hemant
Roll No : 10327296
Abstract:
In the age of Internet, where digital information is rampant, ownership and copyright issues
have inspired researchers to develop robust watermarking techniques for images and videos. In
this work, we present a method of reversible watermarking for images. To embed watermark bits
we use edge pixels present in image. Most of the existing techniques in reversible watermarking
use location identification bits to retrieve the pixels where watermark bits were embedded. Much
of the embedding capacity is usually consumed by location identification bits. Hence total
space available for watermark bits reduces. By using edge pixels for embedding, we are
eliminating the use of any location identification bits. Most of edge pixels have high intensity
values and our human visual system (HVS) is less sensitive to changes in high intensity pixels.
Hence perceptual distortion due to edge pixel pair transformation is also less. We use difference
expansion (DE) to transform pair of edge pixels because transformed pixel’s values are much
closer to their original values. By transforming edge pixels only, we are usually altering only
10% to 15% of total pixels in image. This helps in reducing the perceptual distortion. We test our
method on 65 images, which includes general, medical and military images. Embedding capacity
and amount of data recovered varies depending on the texture of image. We validate the
performance of proposed algorithm in terms of mean squared error (MSE), peak signal-to-noise
ratio (PSNR) and structural similarity (SSIM) index. For medical images relatively better results
are produced.
38
Title : Learning Dynamical Processes Over Graph Via Ssdl
Author(s) : AlatkarPrasad Bapurao
Roll No : 13104011
Abstract:
We consider the problem of prediction of dynamical processes over graphs. Such problems occur
in the context of social networks and in network monitoring. As the graph size increases, it
becomes impractical to take measurements or observations at each node. This thesis proposes a
dictionary learning-based technique that allows interpolation and prediction of such missing
entries. Using the network topology and some training data, a dictionary is built to reconstruct
the underlying process occurring over network. The topology information is incorporated by
utilizing a semi-supervised regularization term that encourages smoothness over the graph. Semi-
supervised learning combined with dictionary learning techniques are used to reconstruct the
underlying process from partial observations. The trained dictionary can predict the process in
online fashion from partial observations solving regularized least squares problem. The learned
dictionary can also help with classifying the network process. The online algorithm proposed for
prediction and classification of processes takes into account both spatial and temporal
correlations of underlying processes.
39
Title : Online Tracking Of Low Rank Matrices From Partial
Observations
Author(s) : Mohan Boda
Roll No : 13104030
Supervisor(s) : Ketan Rajawat
Abstract:
Low rank matrix completion is a problem where a number of missing entries of a matrix are
estimated using only a few observed entries. Such matrices occur in the context of network
monitoring, collaborative ltering, and video processing. There exist e cient o -line algorithms
which are capable of completing large matrices. Popular examples include Singular Value
Thresholding (SVT), Fixed point Continuation Algorithm (FPCA), OPTSPACE, and Geometric
methods such as GROUSE and GRASTA. This work considers the problem of tracking low rank
incomplete matrices that observe many practical scenarios. The proposed methods can be classi
ed into two broad categories. The online methods, such as SVT-LMS, soft-thresholding, and
sub-gradient-based methods can complete slowly changing matrices with high accuracy. These
methods are also cheap, simple, and fast compared to existing state-of-the-art methods. For
highly time-varying matrices, the thesis puts forth a class of smoothing methods that operate
with a constant amount of lag. While the smoothing techniques proposed are more com-
putationally intensive, they yield signi cantly better accuracy, and converge within a few
iterations. The proposed techniques outperform existing techniques when compared over data
collected from network monitoring logs and over video samples.
40
Title : Optimal Encoder Design & Sensor Scheduling in Wireless
Sensor Network
Author(s) : Kumar Sumit
Roll No : 13104147
Abstract:
Wireless sensor networks (WSNs) are spatially distributed systems consisting of several low-
cost, resource-constrained sensors which transmit sense data to a fusion center. WSNs are used
in applications such as environment monitoring, surveillance, search and rescue,disaster relief,
and many more. Detection of weak signal is of primary interest for most of the WSN
applications. This thesis considers the problem of sensor selection for weak signal estimation.
The sensors observe a weak signal, burried in noise, and transmit it to the fusion center. Prior to
transmitting the observations to data fusion center, sensors perform linear precoding on
observations to minimize the mean square error at fusion center. The fusion center receives the
sensor transmissions coherently, and utilizes the MMSE estimator in order to recover the sensed
signal. In order to select the best set of sensors, the log determinant of the estimation error
covariance matrix is utilized. We show that the final objective function is in the form of
difference between two submodular functions. It is shown that the resulting function is
submodular in the set of sensors, and has special properties that can be utilized for sensor
selection. By minimizing objective function we select optimum sensors and show that tuning of
trade-off parameter result in trade between number of sensors selected and estimation accuracy.
41
Title : On Optimal User Pairing and Opportunistic Interference
Alignment in Interference Channels
Author(s) : Sinha Atul Kumar
Roll No : 10327171
Supervisor(s) : Chaturvedi Ajit Kumar
Abstract:
Interference alignment is an advanced interference management technique which exploits

multiple signaling dimensions in wireless networks in order to achieve maximum Degrees-of-
Freedom (DoF). In this thesis, we focus on exploiting user diversity to improve the performance
of interference channels by using interference alignment. We consider a network with K base
stations (BS) and N users (N >= K) in which each BS is allowed to transmit to at most one user.
We consider the scenario where channel state information (CSI) is available at the receivers and
transmitters. Using CSI, we try to determine the BS-user pairing which maximizes the sum-rate.
We propose the optimal user pairing algorithm for maximizing the achievable sum-rate and
demonstrate the gains in sum-rate performance, theoretically as well as through simulations.
Next, we propose a low complexity user pairing algorithm which employs coordinate ascent
approach and achieves sum-rate close to that of the optimal algorithm. Next we relax the
assumptions of channel state information at transmitters and cooperation among them. It is not
possible to attain interference alignment through conventional methods in such scenarios. We
propose a low complexity opportunistic interference alignment (OIA) algorithm which exploits
the available user diversity in order to align the interferences through user selection and is based
on a very low overhead feedback scheme. It is shown that if the number of users in every cell is
scaled with respect to the transmit power, a feasible target sum-DoF can be achieved. Lastly we
extend the user selection based OIA algorithm to the case of user pairing. We illustrate, both,
analytically and through simulations that in a K- transmitter network, user pairing can provide a
K-fold gain over user selection in terms of total number of users required to achieve similar sum-
rate or sum-DoF performance. Similar to the case of OIA with user selection, a feasible target
sum- DoF could be achieved with user scaling.
42
Title : Energy Harvesting Cognitive Radio : Save-Sensing-
Transmit Tradeoffs under Primary User Traffic
Author(s) : Sirigireddy Sudhakar Reddy
Roll No : 13104142
Supervisor(s) : Banerjee Adrish
Abstract:
Spectrum efficiency and energy efficiency are important factors in designing wireless networks.
Cognitive radio technology can improve spectrum efficiency through dynamic spectrum access.
Radio Frequency (RF) energy harvesting is a promising technique to power wireless networks
and thereby improve energy efficiency. Therefore, to attain both spectrum and energy efficiency,
we can incorporate the RF energy harvesting capability in Cognitive Radio Networks. We
consider a self-powered Cognitive Radio system in which Secondary User (SU) has no fixed
power supply (e.g batteries) and extracts energy exclusively via ambient radio signal. Because of
hardware limitations, SU can not perform RF energy harvesting and spectrum sensing
simultaneously. Therefore, we divide the time slot of SU into three non overlapping sub slots
exclusively for energy harvesting, spectrum sensing and data transmission respectively. Here we
consider that the Primary User (PU) is not synchronized with time slot structure of Secondary
User, that means PU changes its status at any time within the time slot of SU. We focus on
maximizing the SU throughput by considering the trade off between three sub slots with random
arrival and departure of PU within the time slot of SU. We formulate the SU throughput
optimization problem in form of Mixed Integer Non Linear Programming (MINLP) and derive
the optimal duration of energy harvesting, spectrum sensing and detection threshold using the
Differential Evolution Algorithm. We present the in-depth numerical analysis how the SU
throughput is affected by the PU traffic, sensing SNR and energy harvesting rate. The throughput
of SU is decreasing with the increase of PU traffic. Throughput of SU is increasing With
increase in energy harvesting rate and with sensing SNR.
43
Title : Voice Activity Detection using Deep Belief Networks by
source and system level information fusion
Author(s) : Arya Devanshu
Roll No : 10327229
Abstract:
Voice Activity Detection(VAD) is an elementary task in all speech related applications such as
speech recognition, speech coding, speaker diarization and speaker verification. Its main aim is
to determine the existence of human speech from an audio signal. In an ideal scenario, this can
be done by simply thresholding energy of signal. However, the most challenging problem of
VAD is to make it perform when the signal is corrupted by noise and the signal-to-noise ratio
(SNR) drops below 10dB. Numerous approaches has been proposed to tackle this problem that
with a variety of feature extraction domains and decision making algorithms. In this thesis, a
noise robust VAD algorithm is proposed using source-filter theory of human speech production.
In general, source based information is not considered in standard VAD methods. However, it is
shown that source based features possess valuable information about speech part of signal even
at very low SNR . The aim of this work is to investigate the joint use of source and system level
features and show the significance of merging long term information to them, in order to develop
a VAD model. Finally, to fuse the advantages of multiple features from different domain for the
robustness of VAD, Deep Belief Network(DBN) based approach is proposed. The DBN based
VAD extracts a new feature that can fully express the advantages of all features by transferring
them through multiple non linear hidden layers. As compared to the existing machine learning
based VADs which only utilize shallow models, the model developed in this work introduces a
deep model for the multiple feature fusion task in VAD. The proposed algorithm uses a lesser
dimensional feature as compared to state-of-the-art DBN model.
44
Title : Joint Source Localization and Separation Using Sparsity
Based Methods
Author(s) : Kalkur Sachin N
Roll No : 13104127
Abstract:
Blind source separation has been an active research topic with an aid to important practical
application such as Acoustic holography and Speech recognition. On the other hand sparse
reconstruction methods for source separation have been extensively studied and applied time-
frequency domain but its study in spatial domain is very limited.\par In this thesis, we propose a
novel joint source localization and separation method using a sparse reconstruction framework.
This task is facilitated by the data model in spherical harmonic domain, eliminating the
dependency between source locations and source strengths, which is generally observed in the
standard data models. Subsequently, the problem is formulated as an optimization problem with
an orthogonality constraint. Necessity of the orthogonal constraint has been described in detail.
The solution to the problem is obtained using splitting method based on bregman iterations. The
sparse vector obtained by solving a nonconvex problem facilitate the process of determining
source locations.\par Experiments are performed on various datasets at different SNRs to
evaluate the performance of the proposed method. The localization results have been evaluated
using RMS error and the separation of sources from the mixture is evaluated using performance
metrics such as LSD, PESQ and SDR. Accordingly, the proposed methods is reasonably better
compared to the standard techniques in the literature. Subsequently, the proposed method
converges in lesser iterations illustrating the fact of lesser complexity. The issue of frequency
dependency has been resolved and optimal frequency selection is presented. Finally in this
thesis, we have highlighted the importance of spatial sampling and spherical harmonic noise and
propose possible future work of source separation with varying harmonic order.
45
Title : Preamble Based Channel Estimation in GFDM
Author(s) : Ghatak Gourab
Roll No : 13104048
Abstract:
Existing preamble based channel estimation techniques give no consideration to the out of band
(OOB) radiation of the transmit preambles. Novel communication schemes like Generalized
Frequency Division Multiplexing (GFDM) require the transmission to have low OOB radiation.
Two particular preamble design techniques are proposed and their performance is analyzed in
terms of OOB radiation and MSE. Improvement schemes are studied to cater to the needs of
practical scenarios in which reduced length preambles are considered and a pinching technique is
used to further reduce the OOB radiation. The obtained preambles are shown to have 5 to 20dB
lower OOB radiation than the existing preamble based estimation techniques. The estimated
channel values are used in equalization of GFDM systems that are aimed for transmit diversity
and their performance is compared to equalization with perfect channel knowledge. The
Minimum Mean Squared Error (MMSE) estimation leads to better results than the Least Squares
(LS) estimation technique.
46
Title : On Redundancy Rate of SWLZ and FDLZ Algorithms and
on Exponential Rates for Generalised AEP and Waiting
Times
Author(s) : Jain Ayush
Roll No : 10327181
Supervisor(s) : Bansal Rakesh K
Abstract:
In the first part of this thesis we obtain a number of results related to redundancy rate of fixed
database Lempel-Ziv (FDLZ) algorithm, sliding window Lempel-Ziv (SWLZ) algorithm and a
variant of FDLZ algorithm by Wyner and Wyner. First we improve the upper bound on the
contribution of pointer bits to the compression ratio for fixed database Lempel-Ziv (FDLZ)
algorithm to be $ H+O(1/\log_2 n) $, from the previous bound of
$H+H(1+o(1))\frac{{\log_{2}\log_{2}n}}{{\log_{2}n}} $ to obtain the improved upper bound
of $H(1+o(1))\frac{{\log_{2}\log_{2}n}}{{\log_{2}n}} $ on redundancy rate, which matches
the existing lower bound to the first order term. Here $H$ denotes the entropy rate of the source
and $n$ is the size of the database. For the variant of FDLZ suggested which uses two databases
to encode the phrase length, we obtain an upper bound on the redundancy rate equal to
$O(1/\log_2 n) $. Next we obtain improved upper bound on redundancy rate for the SWLZ
algorithm. We obtain a pathwise upper bound of $
H(1+o(1))\frac{{\log_{2}\log_{2}n_{w}}}{{\log_{2}n_{w}}}$, on the redundancy rate for this
algorithm, for sources satisfying Markov condition. Here $n_{w}$ denotes the window size. In
the second part of this thesis we analyse the rate of convergence of waiting times $ W_{n}(D) $,
until a $ D- $close version of the first $ n $ symbols of a realization of a process appears in the
realization of another independent process and the rate of convergence in generalized asymptotic
equipartition property (AEP). Statistics of Waiting times $ W_{n}(D) $ plays the key role in
analysis of lossy version of FDLZ algorithm. We first relate rate of convergence of waiting times
$ W_{n}(D) $ with rate of convergence in generalized AEP. We then identify the conditions
under which exponential rates of convergence holds in generalized AEP and for waiting times.
47
Title : Detection and Removal of Cracks from Digitized Artworks
Author(s) : Bansal Ankit
Roll No : 13104015
Abstract:
Ancient paintings and artworks contribute to cultural heritage. Such ancient artworks degrade
due to aging, drying and mechanical factors. With increasing imaging tools, image processing
methodologies are being developed to analyze, preserve and restore the cultural heritage in form
of digitized artworks. Most prominent artifact in these are undesired random patterns called
Cracks or Craquelure. Such cracks degrade the perceived image quality. Various image
processing methodologies are used to detect the cracks and restore the missing part of the
digitized artwork. We propose a method which uses morphological operations to detect the
cracks and other image processing methodologies to reduce the miss detection. Cracks have low
luminance and by making use of this feature of cracks, at first, we preprocess the image to
enhance the contrast to detect the cracks also present in dark regions. After preprocessing, we
apply morphological operations to detect the cracks and smoothen it using a filter. Finally, we
threshold the image to extract the cracks and remove components with area smaller than a user
defi ned parameter. This gives us the fi nal crack map of the image. We compare the crackmaps
of images using proposed method and existing methods. We show the quantitative analysis with
the use of metric to fi nd true detection, false detection and miss detection of crack pixels with
the use of few images. We restore the images by K-SVD based sparse representation. We
compare the results of restoration with other techniques such as simple median filter, modi fied
median filter and exemplar based inpainting.
48
Title : Subspace Based Domain Adaptation for Visual Recognition
Author(s) : Raj Anant
Roll No : 10327086
Abstract:
Domain adaptation techniques aim at adapting a classifier learnt on a source domain to work on
the target domain. Exploiting the subspaces spanned by features of the source and target
domains respectively is one approach that has been investigated towards solving this problem.
These techniques normally assume the existence of a single subspace for the entire source /
target domain. In this work, we consider the hierarchical organization of the data and consider
multiple subspaces for the source and target domain based on the hierarchy. We evaluate
different subspace based domain adaptation techniques under this setting and observe that using
different subspaces based on the hierarchy yields consistent improvement over a non-hierarchical
baseline. Also, in the second part of the work, we propose subspace alignment based domain
adaptation of the state of the art RCNN based object detector [11]. The aim is to be able to
achieve high quality object detection in novel, real world target scenarios without requiring
labels from the target domain. While, unsupervised domain adaptation has been studied in the
case of object classification, for object detection it has been relatively unexplored. In subspace
based domain adaptation for objects, we need access to source and target subspaces for the
bounding box features. The absence of supervision (labels and bounding boxes are absent)
makes the task challenging. In this paper, we show that we can still adapt subspaces that are
localized to the object by obtaining detections from the RCNN detector trained on source and
applied on target. Then we form localized subspaces from the detections and show that subspace
alignment based adaptation between these subspaces yields improved object detection. This
evaluation is done by considering challenging real world datasets of PASCAL VOC as source
and validation set of Microsoft COCO dataset as target for various categories.
49
Title : Musical tempo Estimation using Sub-band Synchrony
Author(s) : Chowdhury Shreyan
Roll No : 10327697
Abstract:
Tempo estimation and onset detection are two important aspects of music information retrieval.
Onset detection aims to locate instances in a music audio where there are note onsets or
percussive hits, while tempo estimation uses inter-onset intervals and other features to estimate
the pace of the musical piece, measured in BPM (beats per minute). Tempo estimation has
applications in music production and mixing, music classification, automatic playlist generation,
and audio-visual synchronization, among other music technology tasks. Numerous methods have
been proposed in literature for tempo estimation with varying accuracies, however, most are
error prone and tend to fail for musical styles that do not have a strong, distinct and steady
percussive beat going on. This thesis proposes three different approaches to address this issue.
The first proposed method uses the fluctuation strength feature to detect the dominant amplitude
modulation frequency in the audio and determines tempo based on the same. The remaining two
methods detect onsets first, followed by estimating the tempo based on the onset curve. The
spectral centroid method detects onsets by calculating the ``center of gravity" of the spectrum at
each time frame, and the sub-band synchrony method detects onsets by locating frames at which
there are coherent changes in the envelopes of different auditory frequency bands. Sub-band
synchrony has been shown to provide the best results for tempo estimation among the algorithms
tested.
50
Title : EMG modeling using Bessel series expansion
Author(s) : Meena Narendra Kumar
Roll No : 10327433
Supervisor(s) : Sircar Pradip
Abstract:
In this thesis, we propose a novel approach to model the electromyography (EMG) signal using
the Fourier-Bessel (FB) series expansion. We show that the FB series can be used to represent
the EMG signal since complete reconstruction of the original non-stationary signal is possible if
the signal is expanded using FB series. Thus, expansion of the signal in the FB series becomes a
way to analyze the signal as well. The FB coefficients can be used to reconstruct the original
signal. The EMG signals used are from tibialis anterior muscle of three subjects with myopathy,
neuropathy and without any history of neuromuscular disease. The quality of the reconstructed
EMG signal was evaluated by computing distortion measures namely the mean squared error
(MSE), Itakura-Saito COSH measure, RMS log spectral measure and relative MSE. The
proposed method has been compared to an existing method where the EMG is represented as a
multicomponent time-varying amplitude and frequency modulated signal. It has been shown that
our analysis technique leads to efficient modeling and data storage.
51
Title : Regularity Flow Inspired Target Tracking in FLIR Imagery
Author(s) : Kumar Nikhil
Roll No : 13104199
Abstract:
In this work, we present a novel and very simple approach of object tracking in FLIR (Forward
Looking Infra-red) imagery utilizing spatio-temporal information of a video sequence. There are
a number of fundamental differences between information content of visible and infra-red
sequences like very poor (Signal to Noise Ratio) SNR, low dynamic range, dynamic nature of
target signatures, dependence upon scene thermodynamics and effect of sun glint which make
job of tracking more tedious in infra-red domain. Generally trackers need an initialization often
done manually or by object detection, but in present approach there is no need to initialize the
tracker. IR signatures of targets are more prominent than background and clutter, this contrast is
commonly used as a clue for detection but in case of poor SNR and small target size detection
based on this fact becomes more challenging. Present approach also relies on this contrast based
− Y plane trajectory
clue but in place of detecting small targets limited in a couple of pixels in X
followed by targets is explored in X− T frame. A small g rou p of frames of v ideo seq u ence is
taken to form a 3D data cuboid with X ,Y ,T axes, this cuboid is re represented as stack of
contiguous X − T slices. As few of these X − T slices contain information related to trajectory of
moving objects, with help of Hough transform based inclined line detection trajectory
information is extracted from slices and used for labelling moving objects − in Y
X plan e. A
dataset of infra-red sequences is generated having targets like tanks, AFVs and other targets of
military importance in presence of substantial amount of clutter in variable atmospheric and
thermodynamic conditions and results obtained demonstrate effectiveness and robustness of
present approach.
52
Title : Hybrid Structured Peer to Peer Live Video Streaming
Protocol
Author(s) : Saxena Abhishek
Roll No : 13104005
Abstract:
A large number of applications on internet require simultaneous delivery of video contents to

large number of users. Recently, peer-to-peer based live video streaming systems have emerged
as a new paradigm to support over 100,000 concurrent users. In this thesis, a Distributed Hash
Table (DHT) algorithm with minimal central control is presented which organises all the newly
arrived peers into a structured overlay network in which binary tree acts as a backbone of the
network and rest all peers are arranged in mesh with these binary tree nodes. This new hybrid
structured peer-to-peer system utilises the advantages of both tree based overlay structure for low
start up delays by pushing the video contents in small packets called chunks along the tree and of
mesh based overlay structure for being less prone to peer churn dynamics by arranging most of
the peers in the mesh part of the proposed network. A BitTorrent like peer-to-peer technology is
used in the mesh part of the proposed network to support the live video streaming. Unlike
BitTorrent, timely and continuous chunk delivery is must in live video streaming systems and
thus a sliding window protocol is used to meet the constraint of playback deadline for each
chunk. The thesis also explores all the key design factors like peer group size, number of
fragments available for sharing and also the ideal fragment size for the optimization of the
proposed network. Analytical study of the delay for the proposed network has also been done
and further it is compared with the push and pull (with no peer-to-peer) methods. Lastly, the
possible avenues for future directions are outlined.
53
Title : Sparsity-Based Source localization under Sensor Position
Errors in Spherical Harmonic Domain
Author(s) : Gupta Avadh Bihari
Roll No : 10327173
Abstract:
Source localization is an important problem which finds wide applications including source
separation, Electroencephalography (EEG), radar systems, and tracking. A large number of non-
parametric and parametric approaches have been proposed for Direction-of-arrival(DOA)
estimation in the past half centuries. Recently a class of semi-parametric algorithms based on
sparse representation(SR) has attracted significant interest due to its capability in achieving high
resolution and dealing with coherent sources but its applications to source localization in the
spherical harmonics domain is very limited. Also, most of the techniques used for DOAs
estimation are developed based on the assumption of exactly known array manifold which leads
to inaccurate data model due to different kinds of perturbation exists in real scenario. In this
thesis, a new sparse representation model is presented which is based on sparse reconstruction
using a spherical microphone array. The source localization method proposed in this work
addresses three important research issues. It formulates the source localization problem in the
spherical harmonics domain as a sparse reconstruction problem. It focus on solving the problem
of sparsity based DOA estimation in the presence of sensor position errors by taking sensor
position errors as normally and uniformly random variables. Subsequently, a low complexity
method to estimate the direction of arrival of multiple sources is also used by using partial
elevation angle dictionaries. Including the statistical characteristics of perturbation and no Taylor
approximation errors being introduced in the proposed SR model guarantee the superiority of the
proposed model in robustness. Experiments like RMSE and Probability of Resolution of sources
are performed on various datasets at different SNR, different source separations and different
array order to evaluate the performance of the proposed model. The results are compared with
the existing methods in terms of robustness and probability of resolution of uncorrelated sources.
The proposed method outperforms the other methods over a long range of SNR.
54
Title : Performance Enhancement with energy efficient Dynamic
Source Routing (DSR) protocol in Mobile Adhoc Network
(MANET)
Author(s) : Singh Nirdesh
Roll No : 13104092
Abstract:
Mobile adhoc network (MANETs) is an infrastructure-less network of mobile nodes that can be
deployed anywhere at any-point of time. In MANETs there is no centralized server that can
provide configuration as well as routing, so each node needs to perform distributively to
establish communication among them.To facilitate communication via multiple hops, each node
will serve the purpose of the host as well as the router for forwarding the data packets in the
network. Each node consumes energy for communication in MANET. To improve network life
time in MANET, nodes should be in sleep mode or in idle mode when they can. The primary
challenge in MANET is that each node needs to continuously maintain the routing information in
order to forward the packets. Due to continuously breaking and establishing routing paths, source
node has to broadcast route request packet again and again. Thus, a suitable routing protocol is
required in MANETs that can improve various network performance metrics. In the thesis, we
propose a scheme to enhance network life time. As a result, source node will select a maximum
energy routing for transmitting data to destination. Maximum energy route will stay for a long
time and overall network performance will be improved. We have simulated with modified
energy efficient DSR in ns-2.35 and compared various parameter with original DSR. We have
consider random way point mobility model so that all the nodes in the network should have same
speed but they are moving randomly to and fro in the network. The final conclusion shows that
modified energy DSR has long network life time and the packet delivery ratio has also improved
as compared to original DSR protocol
55
Title : Anonymity in Delay Tolerant Networks
Author(s) : Vasudevan Ajit Kumar
Roll No : 13104002
Abstract:
Delay Tolerant Networks (DTN) are ideally suited for military applications considering the
conditions of intermittent connectivity between nodes, long variable delays, asymmetric data
rates etc that are common for military environment. Security of data and privacy are of utmost
importance in a military application. In order for DTN to be practical in military, it must be able
to keep the communication anonymous i.e. an adversary should not be able to gain information
on the identity of the sender and receiver of a message. There are several challenges involved in
achieving this in DTN since it is vulnerable to traffic analysis. PKI is not an effective mechanism
for providing security in DTN due to the intermittent nature of these networks and key
management is still an open issue in DTN. Anonymous communication is achieved in TCP/IP
networks like Internet by using applications like TOR which basically provides onion routing
and routes the packets through TOR servers. However onion routing needs the path to be known
beforehand at the sender which is not always possible in DTN. This paper provides an
anonymous communication mechanism - Anonymous Epidemic that provides anonymity in DTN
and uses Combined Public Key (CPK) for key management. CPK is a practical and efficient
mechanism for the conditions of DTN. We evaluate the performance of this routing mechanism
in terms of Delivery Ratio and Average Latency and present the results.
56
Title : Fault Tolerant Clock Synchronization in Distributed
Network using Weighted Average
Author(s) : Sharma P D
Roll No : 13104001
Abstract:
To perform a task in coordinated fashion in distributed environment, we need the common

notion of time. To achieve this purpose, clock synchronization is conceptualized. Clock drifting
is inherent property of a clock, which necessitates the synchronization. A network consists of
nodes which may behave or misbehave. The misbehaving nodes may pose problem in clock
synchronization. We require a mechanism to mitigate the effect of misbehaving nodes. In this
thesis, we present a weighted average clock synchronization algorithm to perform coordinated
activity in a fault tolerant manner. We use the behaviour of the nodes to calculate normalized
weight in localized fashion that lies between 0 to 1. This weight assignment enables us to
suppress effect of misbehaviour up to some extent. This algorithm offers improved precision
while tolerating misbehaving nodes. The upper bound on tolerance limit is one third of the
network size. Our main contribution is to offer improved precision while tolerating misbehaving
nodes.
57
Title : Video Compression Using Video Decomposition Technique
Author(s) : Nellur Naresh Shekhar
Roll No : 13104086
Abstract:
In the present era of digital world, video compression is essential in order to reduce the
requirement of bandwidth in-terms of transmission and storage of videos. All existing video
coding standards developed so far consider video as a sequence of natural frames formed in the
XY-plane, and exploit spatial redundancy in XY-plane and temporal redundancy along T-
direction to achieve compression. New redundancy reduction techniques such as Optimal
Compression Plane have improved the performance of video compression for a number of
coding methods in case of static videos that are captured with little camera motion combined
with small amount of object motion. Compression of dynamic videos with high object motion
combined with complex camera motion is a challenging task. We propose a composite adaptive
preprocessing algorithm which integrates the concept of Optimal Compression Plane and Low-
rank Sparse Decomposition. The proposed algorithm considers video as a 3D data cube, in which
the frames are allowed to be formed in a non-XY plane to exploit the redundancy present in the
video to fuller extent. Low-rank and Sparse Decomposition is performed in adaptively selected
Optimal Decomposition Plane depending on the visual content of the video, to extract the global
background information and reduce the amount of data to be compressed. The main contribution
of this work lies in identifying optimal frame plane for decomposition. We evaluate the
performance of the proposed redundancy reduction technique on video coders such as Motion
JPEG, Motion JPEG 2000, H.264 Intra-only profile, and H.264 for videos with different visual
content and present the results in terms of Rate-Distortion curves.
58
Title : Localization enhancement of wireless sensor networks
Author(s) : Aeddula Narasimha Reddy
Roll No : 13104010
Abstract:
Wireless sensor networks are a significant technology attracting considerable research interest
due to their emerging applications like smart building failure detection and reporting, and target
tracking. In these applications it is necessary to accurately orient the nodes in the sensor network
with respect to a global coordinate system in order to report data that is geographically
meaningful. This thesis aims at improving localization accuracy using different techniques. The
first technique involves the use of mobile nodes anchor nodes for increasing the accuracy of
localization. The mobile nodes is moved to each quadrant in order to improve the location
estimate of other nodes. The second technique considers spectrum sensing for localization of
sensor networks. The goal is to utilize RF cartography, where the spatial distribution of the
power spectral density if estimated using the basis expansion model. The general set up allows
several transceivers willing to cooperate in estimating the location of each other, and uses the
least-squares and lasso formulations.
59
Title : Fast CFO Estimation In OFDM Systems Using Golden
Section Search Algorithm
Author(s) : M Rampriyan
Roll No : 13104115
Supervisor(s) : Vasudevan Kasturi
Abstract:
Orthogonal Frequency Division Multiplexing (OFDM) plays an important role in the

implementation of high data rate communication. The OFDM system transmits data as a set of
low symbol rate parallel data rate streams over orthogonal narrow band sub carriers. This
equivalently converts the frequency selective fading channels to a number of narrow band flat
fading channels thus combating the distortion or the inter-symbol interference of the multipath
channel. However, the OFDM system is highly vulnerable to synchronization errors, mainly the
carrier frequency offset (CFO), which, if left uncompensated results in the loss of orthogonality
of the sub carriers, leading to significant bit error rate (BER) degradation. In this thesis, pilot
symbols are used for the estimation of CFO. We use a golden section search method to estimate
the CFO and compare it with the bisection method. Throughout the thesis we assume that
channel and timing is known perfectly at the receiver.
60
Title : Spatio-Temporal Spectrum Hole Detection with Multiple
Primary Users
Author(s) : Gargelwar Aniket Narendra
Roll No : 13104046
Abstract:
Spectrum sensing is an important area of research in the context of cognitive radio networks
(CRN).We consider multi-antenna cooperative spectrum hole detection in cognitive radio
networks, when there may be multiple primary users which are poison point distributed in space.
A detector based on the spherical test is analyzed in such a scenario. The false alarm and the
detection probabilities, as well as the detection threshold and Receiver Operation Characteristics
are available in closed form. Simulations are provided to verify the accuracy of the derived
results.
61
Title : Preamble based Channel Estimation for an OFDM/OQAM
system
Author(s) : Janapiridi Mahesh Prasad Dora
Roll No : 13104057
Supervisor(s) : Kasturi Vasudevan
Abstract:
Orthogonal Frequency Division Multiplexing/Offset Quadrature Amplitude Modulation

(OFDM/OQAM) is proposed as an alternative to conventional OFDM with cyclic prefix (CP) for
transmission over multi-path fading channels. Two typical features of the OFDM/OQAM
modulation are the absence of a guard interval (GI) and the fact that the orthogonality property
only holds in the real field and for a distortion-free channel. Thus, the classical channel
estimation (CE) methods used for OFDM cannot be directly applied to OFDM/OQAM.
Therefore, we study the transmission of an OFDM/OQAM signal through a time-varying multi-
path channel, where each sub-channel is considered distortionless and the time domain channel
equalization reduced to single-tap channel gain. We derive different CE methods, the first
proposed method only requires the use of a pair of real pilots (POP). In the second method,
denoted as interference approximation method (IAM), we show how the imaginary interference
can be used to improve the CE quality. Several preamble variants of the IAM such as IAM-I
(with imaginary pilots) and IAM-R (with real pilots) are also discussed. In the third method, by
analyzing the noise distribution and correlation characteristics on each subcarrier in
OFDM/OQAM systems we present the maximum likelihood based frequency domain averaging
(ML-FDA) approach for CE. Finally, the performance results obtained for the transmission of an
OFDM/OQAM signal through a standard known channel and Rayleigh fading channel using the
POP method, two variants of IAM and ML-FDA method are compared with CP-OFDM.
62
Title : EXIT Chart Based Performance Analysis Of Fixed Point
Implementation Of Turbo Codes
Author(s) : Sinha Prithwiraj
Roll No : 13104105
Abstract:
Turbo codes have achieved near Shannon limit performance in communication over noisy
channels. The EXtrinsic Information Transfer(EXIT) chart analysis is now an essential part of
turbo code design and an alternative tool to the traditional BER/SER curve to evaluate its
performance. In this thesis we perform EXIT chart analysis of different types of turbo codes
namely, parallel concatenated, serially concatenated and also turbo code consisting of parallel
concatenated non systematic convolutional codes. Moreover, keeping in mind the increasing
necessity and importance of designing low power wireless sensor networks in a variety of civil
and military applications, low power consuming, fixed point implementation of turbo codes has
been considered and analyzed using the EXIT chart.
63
Title : Analytical Model for Information Propagation Speed in
Highway Vehicular Networks
Author(s) : Ranyal Sachin Dev
Roll No : 13104126
Supervisor(s) : SinghYatindra Nath
Abstract:
VANETs (Vehicular Ad hoc Networks) are upcoming wireless network environment for
Intelligent Transportation Systems (ITS). VANETs promise to enhance the road safety and travel
comfort significantly in both highway and city scenarios. Message propagation, either for
emergency or routine purposes, constitutes a major category of VANET applications and is
particularly challenging in infrastructure-less vehicle to vehicle communication scenarios. In this
thesis, we study the information propagation process in mobile ad hoc network formed by the
vehicles which are distributed with exponential statistics on a highway. We derive the analytical
model for the fundamental properties of the information propagation speed (IPS). Using the
model, one can predict the impact on the Information Propagation Speed for various parameters
such as radio range, vehicular traffic density and vehicular speed. The research provides useful
guidelines on the design of vehicular ad hoc networks. Contribution to the work includes,the
modelling of message propagation among vehicles on a highway with a delay tolerant network
setting, taking into account the random distribution of distances between vehicles, the speed of
vehicles and the radio range. The main objective has been to use the vehicles in different lanes
with different velocity, to propagate messages more efficiently.
64
Title : QOS Analysis In Data Network: Stability, Reliability, QOS
Invoke Rate Perspectives
Author(s) : Mishra Pankaj Kumar
Roll No : 13104097
Abstract:
Services on a packet routing network like internet, demand Quality of Service (QoS) for a satisfying
experience. QoS is a guarantee of service. It can be relatively stringent or non stringent. QoS comprises of
bandwidth, delay, delay variation (jitter) and packet loss parameters. Now, there arises a question if QoS
provisioning is always required in a network. The answer is non-affirmative. It depends upon the resource
crunch conditions. One of the criteria is the load condition. As load condition increases, it causes resource
crunch beyond a critical point. Up to this critical point, there is no need to worry about QoS. Provisioning
of QoS becomes important above the critical load condition. Based on the load conditions, a network
operation can be divided in two segments. First when the load conditions are in the range and QoS
support is not required, and second when the load is such that QoS support mechanisms are must for the
delivery of satisfying experience. The work in this thesis is in two parts covers these two segments of
network operation. Firstly, for the situation when QoS provisioning becomes mandatory in a network, this
thesis analyses the QoS issue in terms of two basic parameters: stability and reliability. Secondly, it finds
out the critical load condition up to which QoS provisioning is not required in a network system. Stability
covers the bandwidth and the delay, and reliability deals with the delay and the packet loss parameters. A
network system is stable when number of packets remains bounded in the network while the system runs
for arbitrary long period of time. It is reliable if every packet is delivered across the network in a bounded
time. The stability and reliability contribute towards QoS in an overlapping manner. This work carries out
the stability and reliability analysis of a packet routing network under the Adversarial Queuing Model
(AQM). AQM represents a network system by three elements: the under lying Network Graph, the
Adversary and the Queuing policy. An adversary is a hypothetical entity injecting packets in the network
at some rate r. The rate of adversary determines the load condition. The queuing policy chosen is
SP/FIFO (Strict Priority/First In First Out) scheduling policy, since it resembles the existing QoS
scheduling approach in internet. A multiclass station and multiclass acyclic network with SP/FIFO policy
is designed to bring out the stability and reliability results based on the rate conditions of the adversary.
The work determines the conditions when two class single station network is stable but unreliable and
further generalises it to multiclass. Thesis also derives the stability condition of two class two station
acyclic network and generalises it to multiclass chain of stations. Further, it proposes two policies to
avoid such unreliability in stability preserving manner. Lastly, the thesis work deals with the segment of
network operation upto which resource crunch does not exist and the QoS provisioning is not required.
First question it deals is that, if at all load conditions (rate of adversary) QoS provisioning is needed in the
network system. The answer is no, it is not required till the resource crunch does not occur. Then, what is
the critical load above which QoS provisioning needs to be invoked in a network system? The answer
depends on the network type like cyclic or acyclic. Here, this critical load at which the QoS provisioning
needs to be invoked is derived in terms of adversarial injection rate for an acyclic network based on QoS
parameter of delay. This critical rate of adversary has been termed as QoS Invoke Rate. Thus, the thesis
covers the two segments of the network operations by giving out stability and reliability of acyclic
multiclass networks when resource crunch occurs and a QoS Invoke Rate for an acyclic network for the
range of the network operation where resource crunch is not there.
65
Title : Conversion of Aspect ratios using a novel mirror setup
Author(s) : Gogineni Bhargava
Roll No : Y8127209
Abstract:
Conversion of aspect ratios is sometimes required in signal processing, given the rigidity of
sensor formats, and the high variability of user requirements. Until now, we had, at best, a
certain number of fixed resolutions and aspect rations such as 3:4, 16:9, 16:10, 14:9 etc available
for sensors. Applications had to willy nilly choose one of these and either adapt their
requirements to what was available, or else use the sensor sub-optimally. We propose here a
general design for optical arrangements which, along with a camera in a standard aspect ratio
format, can solve the problem of transforming the cameras fixed aspect ratio into our choice of
aspect ratio. For this, we devise a method/process which configures a simple mirror setup and a
conventional camera pointed towards the mirror setup for obtaining strips of images at different
distances from the camera setup. These images are individually and jointly processed to get a
single (very) wide-angle image.
66
Title : Fault Tolerant Node Disjoint Multipath Routing In Manets
Author(s) : M Venkatesh
Roll No : 13104072
Abstract:
A Mobile Adhoc Network (MANET) represents a collection of wireless mobile devices that
can communicate with each other without the requirement of any pre existing infrastructure via
wireless links. Routing in MANETs is a difficult task since the nodes present in the network are
mobile and the topology of such a network is prone to changes. A Routing protocol which can
adapt to these topology changes is preferred to operate in MANETs. Adhoc On-demand Distance
Vector routing protocol (AODV) is one such routing protocol which establishes a single path
for any given source-destination pair. If the single path between the source and destination
breaks, then the source node needs to reinitiate a new path discovery process. Hence, single path
routing does not support fault tolerance. In MANETs, multiple paths are desired since they can
ensure fault tolerance. The multipaths can guarantee fault tolerance only if they are link disjoint
or node disjoint. Link disjoint paths have no links in common, whereas Node disjoint paths have
no nodes in common. Link disjoint paths ensure fault tolerance with respect to link failures
whereas Node disjoint paths ensure fault tolerance with regard to link failures as well as node
failures. Additionally, Node disjoint paths can also aid in load sharing which can reduce
congestion in the network. Hence, we prefer the formation of node disjoint multipaths for any
given source - destination pair. In this thesis, we have proposed a node disjoint multipath
algorithm based on AODV which can find node disjoint paths between any pair of source and
destination in a single route discovery. We have simulated our proposed algorithm and compared
it with respect to an existing multipath protocol known as Node Disjoint Multi Path- AODV
(NDMP-AODV). Simulation results show that our proposed algorithm finds more number of
node disjoint paths per route discovery cycle and needs lesser number of route discovery cycles
for route maintenance as compared to NDMP-AODV.
67
Title : Papr Reduction In OFDM System By ZADOFF-CHU
Transform
Author(s) : Chaudhry Ravinder K S
Roll No : 13104120
Abstract:
The modern wireless communication era has seen the advent of many new technologies for
meeting the ever increasing demand of high data rates and high performance at low costs.
Orthogonal frequency division multiplexing (OFDM) is a multicarrier modulation technique
popular primarily because of its high-speed data transmission, simplified digital implementation
and effectiveness in combating the frequency selective fading channel. OFDM technique is
widely used in wireless communication nowadays. High peak-to-average power ratio (PAPR) is
one of the major drawbacks for OFDM systems. High PAPR degrades the efficiency of high
power amplifier (HPA). Unfortunately HPA is the most expensive device in the RF module.
According to the literature reduction of PAPR through precoding method (precoding matrix) is
better than using the DFT matrix. This thesis is dedicated to the study of the Zadoff-Chu
Transform for PAPR reduction. The Zadoff-Chu Transform is a symbol scrambling technique for
PAPR reduction, which has a lower computational complexity when compared with other
scrambling techniques such as SLM and PTS.
68
Title : Anthropological Location & Identity Authentication
Author(s) : Pratiher Sawon
Roll No : 13104135
Supervisor(s) : K S Venkatesh&GuptaSumana
Abstract:
In this work, we present different methods of authentication based on concepts borrowed from
cultural anthropology. The methods discussed here explore the applicability of anthropology in
computer systems by focusing on emulating the interactions in our social and cultural sphere by
means of computer algorithms. Most of the existing work on authentication deals with token
based verification methods, conventional shared information based authentication and/or
physiological as well as behavioral biometrics. Our method uses a certain concept used by
cultural anthropologists: This idea holds that conventions and practices followed and specific to
a certain cultural community are completely opaque to non-members of that community. We
apply this idea in our novel construction of an authentication system by treating the authorized
and his computer as a cultural community of two individuals, and the unauthorized user becomes
the outsider. Our authentication system introduces certain other innovations: one is the pro-
vision of location authentication, by means of visual sensors to visually sense the immediate
environment, and compare it against the trained data. This simultaneously verifies the users
identity as well, as the very sequence of execution of the location verification process requires
knowledge shared with the system. Multiple themes have been developed and demonstrated as
examples of how these ideas may be implemented, and to give a picture of the huge range of
options and choices available.
69
Title : Performance Analysis of a Novel Radar CFAR Detector
Author(s) : Yadav Sher Bahadur
Roll No : 13104139
Abstract:
Radar detection procedures involve the comparison of the received signal amplitude to a
threshold. In order to obtain a constant false-alarm rate (CFAR), an adaptive threshold must be
applied reflecting the local clutter situation. Their primary goal is to maintain the desired false
alarm rate and to be invariant to changes in the clutter density function. This is achieved by
adaptively estimating the clutter power based on a finite number of clutter samples within a
processing window. To achieve this, a CFAR detector processes a finite set of range-Doppler
samples within a reference window surrounding the cell under test and sets the threshold
adaptively based on a local estimate of the total noise power. The CA-CFAR detector is optimal
for detecting targets embedded in exponential clutter and noise of unknown power, utilizing
maximum likelihood estimate of the noise power to set the adaptive threshold. The order statistic
(OS) CFAR detector is robust in rejecting impulsive noise and preserving edges. The proposed
novel CFAR detector combine the result of the cell averaging (CA) CFAR and OS-CFAR to get
a better detection performance. In a homogeneous background, the mathematical models of the
two new CFAR detectors are derived and their performance has been evaluated and compared
with that of CA-CFAR and OS-CFAR.
70
Title : Performance of Channel Independent Precoder with and
without Channel estimation for OFDM
Author(s) : Meshram Krunal
Roll No : 13104071
Supervisor(s) : Vasudevan K
Abstract:
Orthogonal Frequency Division Multiplexing (OFDM) provides an efficient means to mitigate

Inter Symbol Interference (ISI) caused by the channel multipath spread and requires simple
frequency domain channel equalization via the Fast Fourier Transform (FFT). However, in
addition to the problem of large peak-to-average power ratio (PAPR), there is a necessity for
accurate frequency synchronization. Channel coding is used to improve the diversity across
frequency and time, and recently linear precoding and block spreading are introduced for OFDM
systems to gain frequency diversity. The precoded data symbol modulated on a subcarrier is now
a linear combination of the original data symbols and if any subcarrier experiences a deep fade,
the original data symbols can still be recovered from the other subcarriers. Hence the system
performance is improved due to increased diversity order. In our thesis, we study the diversity
performance of different channel independent precoders with and without Channel Estimation
for orthogonal frequency division multiplexed (OFDM) systems over fading channels. The
design of precoders is based on the information redistribution of the input modulated symbols
among the output precoded symbols. The proposed precoders decreases the variance of the
instantaneous noise power at the receiver produced by channel variations. The employment of
different interleavers together with precoding matrices allows different configurations of time-
frequency diversity which can be easily adapted to the channel conditions. The size of precoding
matrices does not depend on the number of data carriers in an OFDM symbol. The precoder is
evaluated with a modified Zero Forcing (ZF) equalizer whose maximum gain is constrained by
means of a clipping factor. Thus, the clipping factor limits the noise power transfer in the
receiver de(pre)coding block in low SNR conditions. We also show the performance of precoder
with comb-type pilot based channel estimation with Least Square (LS) and Minimum Mean
Square Error (MMSE) estimates with optimum clipping.
71
Title : Low complexity OFDM based on GIPS-Transform with
Reduced BER and PAPR
Author(s) : Rane Sahil
Roll No : 13104117
Abstract:
The Gaussian Integer Perfect Sequence (GIPS) based transform for orthogonal frequency
division multiplexing (OFDM) system has been proposed in the literature. The transform matrix
is both unitary and circulant, with each column being a perfect Gaussian integer sequence
containing just four non-zero elements of {±1,±j}. This significantly reduces the complexity at
the transmitter. In addition, the dynamic range of the transmitted signal is much less than that of
the traditional OFDM system and hence results in reduced peak-to average-power ratio (PAPR).
However, the bit error rate (BER) remains unchanged compared to conventional precoded
OFDM system. In this work, we have proposed a new random interleaver based system model
using GIPS transform. The proposed approach is found to have a better BER performance than
the GIPS-transform method, when MMSE equalizer is used, without affecting complexity and
PAPR performance. A closed-form expression is derived for the BER in GIPS-OFDM under
frequency-selective fading channels, for ZF and MMSE equalizers. Using the union bound on
BER of linearly precoded OFDM system, we have derived an expression to show that BER
performance of interleaver based system is better than the GIPS-OFDM. Our simulation results
are also in agreement with the derived expression. We have shown the enhancement in the PAPR
performance when partial transmit sequence (PTS) is used along with GIPS-OFDM. Our
complexity analysis shows, that it is more beneficial in terms of complexity, to implement PTS
with GIPS transform than with the conventional OFDM system.
72
Title : Algebraic Independence: Criteria and Structural Results
over Diverse Fields
Author(s) : Pandey Anurag
Roll No : 10327137
Abstract:
We consider the property of algebraic independence of elements over a field. This is a higher
degree generalization of linear independence. Polynomials f1, . . . , fm ∈ F[x1, . . . , xn] are said
to be algebraically dependent over the field F if there exists a non-zero polynomial A ∈ F[y1, . . .
, ym] such that A(f1, . . . , fm) = 0. If no such polynomial exists, we say that f1, . . . , fm are
algebraically independent. We consider the problem of testing whether a given set of
polynomials is algebraically independent. The problem has an efficient (randomized polynomial
time) algorithm based on the Jacobian criterion when the polynomials are given over a field of
zero characteristic. However this criterion fails when the polynomials are over fields of positive
characteristic. The best known algorithm for the positive characteristic case is due to the Witt-
Jacobian criterion which puts the problem in the complexity class NP#P. The thesis aims to find
alternative criteria and algorithms to test algebraic independence of polynomials. We propose a
technique based on polynomial maps and other faithful transformations which in some special
cases, gives a polynomial time algorithm for testing independence over fields of positive
characteristic. We also give an alternative criterion for positive characteristic case based on the
p-adic valuation of the Jacobian determinant. This reduces the problem of testing algebraic
independence to checking if a rational function solution exists to a linear first order partial
differential equation modulo a prime. We further prove using Lüroth’s theorem that two
algebraically dependent polynomials over a field of positive characteristic can be lifted such that
they become dependent over the rationals. This again gives a differential equation based criterion
for testing independence over fields of positive characteristic. We also prove that the minimal
annihilating polynomial of two supersparse polynomials over the rationals is sparse in most of
the cases, giving as well the exact characterization of those cases. We further use this result to
give an alternative randomized polynomial time algorithm for testing independence of two
supersparse polynomials over the rationals. We finally give an efficient higher derivatives based
Jacobian like criterion to test algebraic independence in a special case over F2.
73
Title : Interference Alignment With Channel Uncertainity And
Estimated Channel Through Kalman Filter
Author(s) : Kumar Manoj
Roll No : 13104077
Abstract:
Interference Alignment (IA) is a promising technique to mitigate the multiuser interference

completely but it needs perfect channel state information (CSI) at the transmitter end for
designing of the precoding matrix. In frequency division duplex (FDD) system, CSI at the
receiver is fed-back to the transmitter for use in IA algorithm. Since perfect CSI is practically
unavailable, in this thesis the following aspects of imperfect CSI have been studied: Impact of
additive channel uncertainity and impact of channel estimated by Kalman filter on sum rate.
Abstract Since eliminating the interference from all the users is not necessary because some
user's interference have very less impact on the sum rate due to the large path loss. The number
of users may lead to the signalling overhead and feasibility constraint. To overcome with this
problem, clustering of users into disjoint user is done. Designing of cluster is such that in
between clusters the interference should be very small. In clustered IA system, estimation of
channel is done through Kalman filter and see the changes in sum rate.
74
Title : Precoded Superimposed Pilot For Channel Estimation In
Rayleigh And Rician Fading Channels: Uplink Massive
Mimo
Author(s) : Jain Anmol
Roll No : 13104017
Abstract:
Massive MIMO (multiple-input multiple-output) is a multiuser MIMO technology where each

base station (BS) is equipped with an array of large number of antennas. For effective operation
in multicell multiuser Massive MIMO system, BS requires accurate channel state information.
But large numbers of users and limited coherence time introduces problem of pilot
contamination which deteriorates channel estimation. This thesis introduces precoded
superimposed pilot scheme for channel estimation and data detection in uplink transmission. We
derive expressions for mean square error, self-interference, cross-interference and sum-rate in
both Rayleigh and Rician fading channels. Simulation results confirm that the proposed scheme
reduces self-interference and increases sum-rate compared to conventional superimposed pilot
scheme.
75
Title : Explorations Of Multidimensional Modulo Tic Tac Toe
Author(s) : Tiwari Asutosh
Roll No : Y9227152
Abstract:
In this Thesis,we would introduce readers to variants of the common game Tic
TacToe,includingModuloTicTacToeandReverseModuloTicTacToe. In
anormalTicTacToegameplayedona3x3boardaconsecutivesetofthree
pointsisconsideredtobeasuccess.Fora3x3board,n=3andk=2.Wewillbe discussing
totalpossiblenumberofsuccessesfordi erentsetsof(n,k).Wewould see
possiblerelationshipsbetweennandkandtotalnumberofsuccesses.Wewould
beobservingsamerelationshipsforModuloTicTacToe.Wewouldobservesame cases
forReverseModuloTicTacToeaswell.
76
Title : Shape reconstruction for ISAR imaging
Author(s) : Papanna Jayaprakash Koppa
Roll No : 13104058
Supervisor(s) : Naik Naren
Abstract:
Inverse Synthetic Aperture Radar (ISAR) imaging is a technique used for reconstructing radar
images of targets. Modern high resolution radars are equipped with the hardware needed for
performing ISAR imaging. ISAR images are obtained by signal processing carried out both
online/offl ine are a graphical representation of the target reflectivity function. The use of radar
images will aid Automatic Target Recognition (ATR) systems in identi cation and classi cation
of the targets of interest. The performance of ATR systems can be improved by using radar
images of the target. In the present work, the problem of direct shape reconstruction of an object
using ISAR data is considered. At first, the linear-static case is considered in which all the
scattering centers of the objects are estimated and then the shape is reconstructed. The problem
of direct shape reconstruction is considered next and the problem is modeled as nonlinear least
squares problem. The boundary of the object is represented using closed cubic B- spline curve,
whose control points can be changed individually. The object is modeled using the concept of
zero level sets representation and signed distance function. The Frechet derivatives are computed
and veri ed as part of the minimization of the nonlinear objective function. A nonlinear-iterative
reconstruction algorithm is applied on ISAR data. Levenberg-Marquardt algorithm is
implemented to solve the underlying non linear least squares problem and the unknown control
points are estimated. For, the linear static case , the reconstruction is achieved for an aircraft type
shape with noise added in the measurement data. The output is demonstrated via numerical
simulation. For the, non linear static case, the Frechet derivative is computed and verifi ed with
the finite diff erence method and preliminary reconstructions of basic shapes are achieved. The
proposed algorithm serves as an aid in automating the process of target recognition in military
systems.
77
Title : Marker Less Hand Detection And Gesture Recognition
From 2d Sensor Data
Author(s) : Bolluru Praveen
Roll No : 13104031
Supervisor(s) : K S Venkatesh
Abstract:
Hand gesture recognition systems can be used for human-computer interaction (HCI). The use of
hand gestures provides an attractive alternative to cumbersome interface devices for HCI. Proper
hand segmentation from the background and other body parts of the video is the primary
requirement for the design of a hand-gesture based application. This thesis discusses about
continuous hand gesture recognition. It reports hand detection as well as segmentation
algorithms which are based on skin colour estimated from the face of the human agent i.e. free
hand detection.The face of human is detected using Viola/Jones Face Detector. We used face
detector not only to remove the face also we estimated skin colour range for the human agent. In
our approach, video frames are captured from a low cost webcam (camera) for vision based
gesture recognition technique. The features of hand skin color in two different colour spaces, viz.
the YCbCr and HSV color space are jointly used to detect the hand. Further, the fingertips of the
hand are used to form a distinctive constellation of points, which is highly specific to the human
hand and cannot be easily encountered in other objects. This constellation yields the position and
orientation of the hand and fingertips, thus enabling us to design various gestures. The algorithm
has been used to design a few specific gestures. This involves tracking the change in position of
the hand or change in its orientation. Gestures have also been designed for two hands used
simultaneously, keeping a track of their positions with respect to each other. The system works
in real time. The software used for the project is Computer Vision in Microsoft Visual Studio
using C and C++
78
Title : Estiimation and Cancellation of Lens Flare
Author(s) : Tundulwar Aditya
Roll No : 13104008
Abstract:
In this work, we focus on one of the most common disturbances in images, lens flare or
unwanted light scattering or reflection inside the camera lens system. This stray light in an image
is unwanted in professional photography, though some professionals have become used to
treating it as an aesthetic factor. Recently developed camera lens systems are more prone to this
effect as they use combinations consisting of multiple lenses. We also explain how reflection
lens flare appears around the flare line, particularly joining optical center and light source.
Positional symmetry between lens flare position and light source is explained with linear
regression model. The radial symmetry of camera lens system is taken into consideration to
estimate the lens flare for a captured image. We also have shown that important properties of
camera lens system can be found out using reflection lens flare such as optical center of camera
and shape of aperture of camera. Location of optical center in captured image also gives some
information about lens system of camera and sensor plane of camera i.e. CCD array.
79
Title : Precoding with time-domain feedback in MIMO-OFDM
systems
Author(s) : C Laxmikanth Rao
Roll No : 13104032
Abstract:
Precoding exploits channel state information (CSI) at the transmitter to enhance the performance
of the system. For most practical systems, channel state information is not readily available.
Hence, CSI is conveyed through a feedback link from the receiver to the transmitter. In such
systems, limited feedback resources, associated feedback delays, reduce the usefulness of
channel state information at transmitter (CSIT). Aiming at reducing the feedback overhead,
vector codebooks that are known both to the receiver and transmitter are designed. The receiver
estimates the channel and chooses a precoder matrix based on a performance criteria from a _nite
set, and sends the index of the matrix to the transmitter using limited number of bits. For MIMO-
OFDM systems, precoder matrices have to be designed for every subcarrier increasing the
feedback overhead linearly with the number of subcarriers. To overcome this clustering,
interpolation schemes are proposed which exploits the correlation of adjacent subchannels in
OFDM. In this work, we show for slowly time varying channels, feedback of quantized channel
taps to the transmitter performs better than limited feedback precoding for a given feedback
overhead. BER simulations validate the performance of limited feedback precoding and time
domain quantized channel feedback precoding with the linear MMSE receiver.
80
Title : On History Assisted Spectrum Sensing for Cognitive Radio
Author(s) : Darmwal Yogesh
Roll No : 13104162
Abstract:
Cognitive radio(CR) is an intelligent communication system that can solve spectrum scarcity
problem. It allows secondary user to access underutilized radio spectrum in non-interfering
manner. To access the spectrum, cognitive radio needs to sense its RF -environment to know
whether the band is free or not. Therefore spectrum sensing is an important task of CR. A
number of spectrum sensing techniques have been proposed in the literature to find out the
spectrum opportunities in the temporal domain. However most of them are periodic in nature. A
periodic spectrum sensing technique is one which sense the channel periodically after every τ
seconds. With the help of learning capability of cognitive radio, an aperiodic spectrum sensing
technique can be proposed. In aperiodic sensing technique, cognitive radio learns the pattern of
primary user occupancy behavior and based on the learned behavior, it can predict future
behavior of primary user. We need to sense the channel only when we are not confident about
the state of channel. Thus aperiodic spectrum sensing improves spectrum utilization by
minimizing wastage of resources. In this work, we present a new spectrum sensing technique, in
which prediction of future state of PU (primary user) is used to identify time instances where
sensing is required and for all other time instances we use prediction results to know the state of
the channel.We show through simulations that the proposed method method works better than
periodic spectrum sensing.
81
Title : On Techniques to Detect Malicious Users in Cooperative
Spectrum Sensing
Author(s) : Patra Swarup Suman
Roll No : 13104148
Abstract:
Increasing use of wireless technology has lead to greater demand for radio spectrum. As
spectrum is a scarce resource, it needs to be regulated and available radio spectrum usage should
be optimized. Cognitive radio is an intelligent wireless system that opportunistically access the
channel and increases the efficiency of spectrum utilization. However, the cognitive radio’s
spectrum sensing performance degrades severely when few malicious users are present. The
malicious users report false information to the fusion center thereby causing the fusion center to
make a false spectrum sensing decision. In this thesis we propose a proximity based outlier
detection method known as k-nearest neighbor approach, that does not require an estimate of the
number of outliers. We have also considered the effect of secondary users (cognitive users)
position relative to primary users (licensed users) on system’s performance. We have also
proposed a goodness-of-fit based cooperative spectrum sensing method using Anderson-Darling
test (AD) and Kolmogorov-Smirnov (KS) test. This method uses the empirical CDF of the
received energy values and compare it with the theoretical distribution of each hypothesis. Then
we use Dempster-Shafer theory (DS) to combine the results of both AD test and KS test.
Proposed tests outperform tests based on single goodness of fit tests.
82
Title : Image Binarization of Historical Degraded Document
Images
Author(s) : Chirag
Roll No : 13104033
Supervisor(s) : GuptaSumana&Venkatesh K S
Abstract:
Document image binarization in the presence of di erent types of degradations is a challenging

task.Binarization is the preprocessing step required to separate out text region from non-text
region in a document.It is an important step in document analysis as it a ects the performance of
Optical Character Recognition(OCR). In this thesis we present a novel method of binarization of
historical document images a ected by typical degradations such as stain,ink-bleeding and non-
uniform background.The method can handle document images a ected by multiple defects.The
method is based on choice of appropriate thresholding techniques.The proposed method is tested
on di erent degraded data sets and its performance is compared with existing methods in terms of
F-Measure, Misclassi cation Penalty Metric(MPM) and Peak Signal to Noise Ratio(PSNR).
83
Title : Utilisation of heat generated in high concentration 2D
photovoltaic systems by thermoelectric converters
Author(s) : Goswami Bhanprakash
Roll No : 13104027
Supervisor(s) : AnandRaghubir Singh&Kumar Iyer S Sunder
Abstract:
Concentrated Photovoltaic technology has potential to produce electricity efficiently and cost
effectively. As a part of this thesis work, a 1.8 kWp high-concentration 2D photovoltaic system
(CPV) located in the Solar Energy Research Enclave (SERE), IIT Kanpur has been studied with
an aim to improve its performance. For a start, the devices were spruced up by placing and
ensuring that the triple-junction solar cells at the focus of the concentrators. The connecting
wires were shielded by aluminium sheets so that wires were protected from an off focus beams
between the time the system moves to face the Sun. Due to resource constraints, other
improvements were not carried out for the time being. All the same, the efficiency of the system
increased to about 11% from less than 5% at the start of the project. Due to the high
concentration (1000 X) of solar irradiation in the CPV systems, a large amount of heat is
dissipated in the solar cells. This heat is supposed to be transferred to the environment by natural
convection using heat sinks or by active cooling with the help of fluid (water) circulation. As a
part of this thesis work, the heat developed in the CPV systems was converted to the electricity
using thermoelectric converters. The CPV system and the thermoelectric generator were studied
independently and in an integrated form for their voltage-current-temperature characteristics. To
dissipate heat and create temperature difference on two sides of the thermoelectric generator,
various topologies using static and forced air and water have been tested. As expected, forced
water created larger temperature difference and give better thermoelectric generation. To utilise
the heat generated in the CPV system, two designs were explored. In the first design, the
thermoelectric generator (TEG) was placed inside the CPV module between the solar cell and the
heat sink. In the second design the TEG was placed on the metallic body at the back of the CPV
module. It was found that, although first design resulted in better thermoelectric generation, due
to low heat transfer through the TEG, the temperature at the solar cell increased more and
deteriorates its PV performance. The overall efficiency when using this design actually
decreased to 9.69% for air cooling and to only 10.79% with water cooling. In the second design,
thermoelectric generation was found less compared to the first design, but in this design CPV
power generation was not getting affected as much. The overall efficiency marginally increased
to 11.04% for air cooling and 11.12% for water cooling with this design. Further optimisation of
design should help extract more power by combining TEG and the CPV.
84
Title : Thermal Resistance modeling in FINFET and FDSOI
transistors
Author(s) : Kompala Bala Krishna
Roll No : 13104069
Supervisor(s) : Chauhan Yogesh Singh
Abstract:
While device dimensions continue to shrink down to the 16-nm node, self-heating e ects emerge
as a pressing problem, detrimental to leakage current and mobility in both Fully Depleted Silicon
On Insulator (FDSOI) transistors and Fin Field E ect Transistors (FinFET). This work examines
the heat dissipation paths in the context of nanoscale transistors and propose models for thermal
resistance variation in both FDSOI and FinFET transistors. In this work, 3-D FDSOI device
simulations with variation in both channel length and channel width are done using Sentaurus
Technology Computer Aided Design (TCAD). From these 3-D device simulations, we see that
thermal resistance increases with reduction in channel length. A compact model for
geometrically scalable thermal resistance in FDSOI transistors is developed after extensive
analysis of 3-D TCAD simulation data at di erent ambient temperatures. This model is tested
with industry standard BSIM-IMG model and shows excellent agreement with the experimental
data. In addition, this work also includes the simulations of 3-D FinFET device with variation in
the number of ns (Nfin), shape of ns and n pitch (Fpitch) using Sen- tuarus Process. It is
investigated that thermal resistance has nonlinear dependency on Nfin and Fpitch. A model for
thermal resistance behavior correctly with Nfin and Fpitch variation is proposed. The proposed
model is tested with industry standard BSIM-CMG model and has shown good agreement with
the experimental data and TCAD simulations.
85
Title : Simulatioin And Modelling Of Organic Light Emitting
Diode
Author(s) : Saraba Epili Raja Kirana Saraba
Roll No : 13104044
Supervisor(s) : Mazhari Baquer
Abstract:
Organic light-emitting diodes (OLEDs) are being actively developed for flat panel active matrix
displays and solid state lighting due to their unique properties such as simple fabrication process,
wide viewing angle, thin and flexible form factors and potential low cost. The process of light
emission is based on injection and recombination of electrons and holes from electrodes into a
stack of organic layers under the influence of applied voltage. Of the various factors, electron
and hole current balance is prerequisite to achieve OLED efficiency. Due to widely different
electron and hole mobility in organic semiconductors, current balance is normally poor and has
to be improved through use of specialized techniques. In this thesis, two techniques namely use
of a bi-layer OLED structure and addition of dopants that act as a trap for one carrier are
investigated in detail. In a bi-layer OLED, energy offset at the heterostructure plays a key role in
achieving current balance. To investigate the impact of energy offset on current transport, current
in a hole-only device is studied in detail. An analytical model is developed which correctly
predicts an injection limited current behavior with a cubic dependence on both voltage and layer
thicknesses and an exponential dependence on hole barrier at organic-organic interface. The use
of dopants acting as traps for the more mobile carrier was investigated in detail. It is shown that
through judicious choice of trap energy and concentration, mobility of one carrier can be
selectively decreased while keeping mobility of other carrier unchanged. Current balance and a
broad recombination zone spread over the entire device could be achieved. However, the
characteristics were sensitive to trap characteristics and initial mobility values.
86
Title : Study of suitability of four di-keto-pyrrolo-pyrrole
derivatives for active layers in organic solar cell
Author(s) : Saurav Kumar
Roll No : 10327363
Supervisor(s) : Iyer S Sundar Kumar
Abstract:
Development of high efficiency organic solar cells is an actively pursued area in current
research. This requires synthesis of new materials which have enhanced mobility and high
absorption coefficient in visible region. Di-keto-pyrrolo-pyrrole is being investigated in this
direction because of its good electrical and optical properties. This research work studies the
applicability of four molecules based on Di-keto-pyrrolo-pyrrole derivatives synthesized by Dr.
Somanathan's group at CLRI, Chennai. These molecules have a broad spectrum with high
absorption coefficient in visible region which makes them promising for solar cell applications.
Also, these materials have band gap around 1.5eV which is very close to Shockley–Queisser
optimal band gap for maximum attainable efficiency of solar cells. Experiments for mobility
measurements were performed by making hole only devices with PEDOT:PSS and gold as
injecting electrodes. Single layer devices were fabricated by spin coating using chloroform as
solvent. Mobility of these materials was observed to be of the order of 10-5 cm2 V-1s-1. It was
noted that addition of small amount of diiodooctane to the solvent improved device consistency
and increased the mobility by a factor of almost 5. Finally, two types of solar cells were made,
first having bulk-heterojunction architecture and another with bilayer architecture. Bilayer
devices were recorded to be more efficient for some molecules, while bulk-heterojunction
structure showed more efficiency for others. Efficiency of these molecules was increased by
orders of magnitude upon addition of diiodooctane in chloroform in case of bulk-heterojunction
solar cells, however, bilayer cells had mixed behavior. During the course of experimental work
done for this thesis, the peak efficiency registered was 0.2% in bulk-heterojunction solar cell
with one of the four molecules being worked upon
87
Title : Extraction and modeling of threshold voltage in organic
thin film transistors
Author(s) : Tiwari Swati
Roll No : 10327756
Abstract:
The field of organic electronics has been drawing attention from researchers around the world for
many years. Organic thin film transistors (OTFTs) have potential for development of active
matrix displays, RFID tags, sensors and various other applications. To realize large scale organic
electronics circuits, availability of a compact model of an OTFT becomes a key requirement. The
rigorousness of compact modeling strongly relies on input parameters such as charge carrier
mobility, contact resistance and threshold voltage. Threshold voltage is very important parameter
for device modeling and circuit design using OTFTs. In this thesis work, threshold voltage
parameter of OTFTs has been studied in detail. We have compared five different methods for
$V_T$ extraction on transfer characteristics of an OTFT, simulated on a Silvaco atlas. These five
methods are Extrapolation in Linear Region (ELR), GMLE, Ratio method, VX method and GV
method. We have analyzed the effect of including gate field dependent mobility model, interface
traps and parasitic series resistance present in OTFTs on all above mentioned $V_T$ extraction
methods. We have also developed an analytical expression for threshold voltage based on
assessment of surface potential for accumulation mode OTFTs. Our model provides a way to
calculate surface potential based on the accumulated charges at semiconductor-insulator
interface. We have used the basic fact that accumulation charge density follows exponential
dependence on applied gate bias in sub-threshold regime and power law at higher biases. The
results from the analytical expression are in close agreement with simulations. The simulated
results are obtained by applying ELR and VX methods on I(DS)-V(GS) characteristics
88
Title : Analysis of photo-response of top-contact OTFT
Author(s) : Kumari Smriti
Roll No : 10327718
Abstract:
Organic thin film devices are being actively developed due to the potential for implementing low
cost flexible electronic systems using mass printing techniques. An organic photo-transistor
combines the switching and amplifying property of a transistor with the light sensing capabilities
of a photo detector thereby offering a device that can exhibit higher sensitivity and benefit from a
simplified fabrication process. Although, the optical response of organic thin film transistors has
been widely reported, there has been no systematic investigation of the detailed mechanism
underlying the photo-response in these devices. This thesis uses 2D device simulations to
investigate in detail the different possible mechanisms that can give rise to photo-sensitivity in a
top contact organic thin film transistor. For devices, consisting of a single active semiconductor
layer, the breakup of photo-generated excitons constitutes a key bottleneck especially when
applied voltage is low in the linear mode of operation. In this case electric fields present are
insufficient to break the excitons and photo-response is poor. In the saturation mode, when
applied drain voltage is large and good source injecting contacts are used, a large field forms
within the pinch-off region next to drain and exciton breakup occurs and photo-response is
strong. It is shown that current under illumination increases due to effective modulation of hole
channel length (p-type device) and formation of an electron channel. Electron channel can also
be formed in dark by aligning work-function of drain electrode with the LUMO of semi-
conductor. It was also noted that upon illumination current is not limited near oxide-
semiconductor interface but starts spreading in the bulk. The effects of changing electron
mobility, excitonic parameters and applied voltages are reported in detail. It was found that lower
magnitude of gate voltage is better suited if one desires high sensitivity towards light.
89
Title : Study of suitability of four imidazolin-5-one derivatives for
active layers in organic solar cells
Author(s) : Gupta Raghav
Roll No : 10327553
Abstract:
New organic compounds are being actively developed for applications in solar cells owing to the
high absorption coefficients and low processing costs. In this respect, the derivatives of the main
chromophore of the highly fluorescent "Green and Red Fluorescent Proteins (GFP \& RFP)",
imidazolin-5-ones, appear promising candidates owing to their excellent optical and electrical
properties. This thesis investigates four such derivatives of imidazolin-5-one for organic solar
cell applications. Wet processing with chloroform as the main solvent was used as the primary
method to fabricate thin films and devices as opposed to most of the previous work done on
imidazolin-5-ones where thermal evaporation was used. Absorption studies revealed that these
molecules have high absorption coefficients in the blue-green region of the spectrum.
Profilometry studies were conducted to study the roughness of the films. The electrical properties
of the molecules were investigated by fabricating hole only devices with gold as the cathode. It
was found that the molecules have mobility values that are orders of magnitude higher those of
the previously reported molecules 2.1 x 10-4 cm2 V-1 s-1 as compared to 2.9 10-7 cm2 V-1 s-1
reported previously. Finally single layer devices and bulk heterojunction solar cells (with PCBM
acceptor) were fabricated to study the photo-voltaic effect in the molecules. The device structure
was optimized by testing two solvent combinations and two different hole transport layers. Peak
efficiency of 0.024% was obtained in (Z)-4-(4-methoxybenzylidene)-1-methyl-2-((1E,3E)-4-
phenylbuta-1,3-dienyl)-1H-imidazol-5(4H)-one (M2)-PCBM (1:1) bulk heterojunction solar
cells of with PEDOT:PSS as the hole transport Layer.
90
Title : Effect of annealing poly 3-hexylthiophene layers in the
presence of pulsed electric field
Author(s) : Gangwar Anshik
Roll No : 10327119
Abstract:
Organic solar cells (OSC) have shown the potential to feed the ceaselessly soaring power
demands of our society. The processing steps for fabrication of OSCs are much simpler, cost
effective and energy efficient compared to conventional silicon solar cells. Hence, a lot of
research interest is being taken to improve the performance of OSCs. Poly(3-hexylthiophene-2,5-
diyl) (P3HT), now a days, is one of the most widely used donor material in OSCs. The thesis
presents a comparative study of morphological and electronic properties of P3HT films annealed
in the presence of pulsed electric field of 2000 Vcm-1 at several duty cycles and different
frequencies. For comparison, conventionally annealed films and films annealed in the presence
of a constant electric field of 2000 Vcm-1 were used as reference samples. X-ray diffraction
(XRD) measurements were performed to understand the variation in crystal sizes of films.
Analysis of XRD data confirms the increase in crystal size with the increase in duty cycle for
each frequency. Current density-voltage (J - V) data in the dark for single layer P3HT devices
was fitted to field dependent mobility J - V model at higher biases in space charge limited
current region to estimate zero field mobility (μ0) of P3HT films annealed in the presence of
different electric fields. The extracted mobility follows the same pattern as observed for crystal
size. More than 15% increase in zero bias mobility is observed with pulsed electric field at 90%
duty cycle and 100mHz frequency than the conventionally annealed reference samples.
91
Title : Oscillator And Amplifier Circuits Based On Single Layer
Organic Devices
Author(s) : Gajbhare Sandeep Sanjayrao
Roll No : 13104045
Abstract:
The field of organic semiconductor based electronics has seen significant and unprecedented
progress in the past decade. Low-cost, less energy-intensive and high-throughput production,
implementation on flexible and non-planar surfaces, novel applications, as well as the potential
to move to more environmentally friendly electronics makes this technology particularly
attractive. A wide range of applications of organic electronics are being currently explored,
including displays, lighting, solar cells, printed RFID tags and disposable low-cost sensors. In
this work also we will explore the use of the organic single layer device in many oscillators and
analog amplifier. The simplest organic devices using pentacene as single layer of semiconductor
were fabricated and characterized to create a device model. The oscillator and amplifier circuits
were designed with the replacement of the linear passive resistors with the non-linear single layer
organic device models and the extracted results from these circuits are compared with the
conventional resistor based oscillator and amplifier. The effect of the non-linearity of I-V of the
single layer organic device on stability and output waveforms of the oscillators were analyzed.
Whereas in common emitter amplifier the effect of non-linearity of device is analyzed in terms
of change in amplifier gain, harmonic distortion, etc. The aim is to replace the resistor with the
single layer organic device and analyze the behavior of the analog circuits.
92
Title : Compact Modelling Approach For Organic Thin Film
Transistors
Author(s) : Upadhyay Aditya
Roll No : 13104009
Abstract:
Organic Thin film transistor (TFT) based flexible and printable electronics is a promising area.
Recent advances in fabrication process for Organic TFTs have shown consistent improvement
over the years. Accurate and physical simulations of Organic TFTs are critical for evaluating the
performance of discrete devices and design of complex circuits. TCAD simulation provide
valuable insight into the physics and operation of Organic TFT, however simulations of large
circuits based on Organic TFTs are very expensive and slow in TCAD. Compact model with
sufficient accuracy is required for simulation of large circuits based on Organic TFTs. This
Thesis presents a modelling approach for Organic TFT by including physical effects like gate
voltage dependence of mobility and exponential density of states and defects in a surface
potential based compact model. Computational efficiency is achieved by developing analytical
expression for surface potential. Model’s performance is compared with the data from ATLAS,
SILVACO simulations and a good match is achieved in simulated results.
93
Title : Time Dependent Degradation Study of Performance of
P3HT:PC60BM and PTB7:PC70BM Based Organic Solar
Cell Devices
Author(s) : Yadav Jyoti
Roll No : 13104062
Abstract:
Solar energy is a very compelling source for fulfilling the energy requirements of our planet.
There are many PV technologies such as Silicon based solar cells, GaAs based solar cells which
are commercially produced but their complete potential has remained unfulfilled due to higher
cost of electricity produced by these panels. Among new technologies, organic solar cells have
potential to succeed as a future PV technology due to its potentially environment friendly nature
and low cost of materials involved. Moreover, these devices offer advantages of flexibility and
conformability on a wide variety of surfaces and hence offering a huge application domain.
However, poor stability of organic solar cells in ambient conditions is a big impediment to
achieve these goals. Moreover, analysis of stability requires time dependent studies. Stability of
organic solar cells can be improved by use of encapsulation methods and the conventional one is
to use the glass based encapsulation. Whilst this method provides excellent impermeability to the
diffusion of oxygen and moisture into the device, high costs and rigidity associated with glass is
in contrast to the purpose of organic solar cells in many ways, in particular flexibility and low
cost using roll to roll processing methods. Hence it would be desirable to develop a thin film
based encapsulation which can be flexible as well as should be possible to be integrated with
device processing whilst providing protection equivalent to glass. In this work, we have carried
fabrication and characterization of two bulk heterojunction devices: one based on the blend of
well established material system P3HT:PC60BM and another using PTB7:PC70BM. We have
carried out time dependent degradation studies on these devices and our results suggested that
the device based on PTB7:PC70BM degrade much quicker than the devices using
P3HT:PC60BM. We further carried our studies on P3HT:PC60BM by encapsulating them with a
thin film of thermally evaporable oxide MoO3 using various combinations. However results
suggested that MoO3 does not improve the device stability and that there is a need to investigate
more materials which can be used as thin film encapsulant.
94
Title : A Truly Passive Pixel Image Sensor Array
Author(s) : Rituraj Rituraj
Roll No : 10327605
Abstract:
Photodiodes are the most common choice for image sensing applications. In a large sensor
matrix generally transistors are also required along with the photodiode in each pixel. The basic
purpose of the transistor is to act as a switch to reduce crosstalk or interference from other
unselected pixels. In an active matrix sensor network, the pixel circuitry could be used for
various other operations besides switching. The pixel circuit can be a simple source follower
amplifier in a 3-T active pixel sensor or as complex as an ADC (Analog to Digital Converter) in
a digital pixel sensor requiring tens of transistors. These added functionalities come at a cost.
The fill factor (fraction of total pixel area available for photodiodes) becomes small. Also the
device fabrication becomes more complex increasing the processing cost. This might be
necessary for high end applications but for many other noise insensitive applications a simple
passive matrix sensor is the cost effective solution. Through this work we propose and
demonstrate through simulations and experiment two novel photo sensor devices having inherent
switching characteristics when biased in a certain region and thus making the use of in-pixel
transistor for switching unnecessary. The resulting sensor matrix has a very simple structure,
ideal fill factor, could achieve high spatial resolution and is very easy to fabricate. We have
realized the device using organic materials which are known to have superior optical properties.
This also has an added advantage of being low cost, requires simple processing steps and could
be done on flexible substrates as well. Besides its use as a camera, the device could be used for
developing a cheap, flexible and portable image scanner as well. The large area scanner would
no longer require mechanically moving parts for scanning as is the case with conventional
scanners.
95
Title : Effect of processing asymmetries on a-IGZO TFT
characteristics and OLED pixel driver circuits
Author(s) : Chaudhary Kavita
Roll No : 13104066
Abstract:
Indium-Gallium-Zinc-Oxide (IGZO) thin film transistor is one of the most popular area of
research due to its higher mobility in comparison to convectional silicon thin film transistor.
Now-a-days most of the industries are using inverted-staggered bottom gated etch-stop (ES) TFT
structure of -IGZO. In this structure, generally there is symmetrical overlap between source/drain
contact and gate contact. But these TFTs has very small channel length about 5 m to 10 m, due to
which during fabrication source and drain overlap may slightly differ from their exact position
and result in unsymmetrical overlap structure. These source/drain overlap asymmetries affect the
electron path and consequently changing the electrical properties of -IGZO TFT for example
parasitic resistance, drain current, threshold voltage etc. Currently -IGZO TFTs are used as
backplane in display technologies in both OLED (organic light emitting diode) and LCD (Liquid
Crystal displays) pixel driver circuit. Change in properties of the TFT device will also affect the
operation of OLED. It is necessary to understand the characteristics of these TFTs thus simulate
the -IGZO TFT with different overlap errors which occur during the fabrication of -IGZO TFT,
and model the effective length and parasitic resistance and other electrical properties such as
threshold voltage and sub-threshold swing. In this thesis -IGZO TFT device with same channel
length but varying source/drain overlap is studied to investigate the exact effect of overlap on
electron path and channel of the TFT device. The simulated results are further examined for
device electrical properties. Later the spice simulation of AMOLED pixel circuit is done to
analyze the operation of OLED pixel circuit like charging, discharging and switching.
96
Title : Correlation between Dark & Light Characteristics of Bulk
Hetero-Junction Organic Photo-Voltaic Cells
Author(s) : Imamuddin Mohammad
Roll No : 13104079
Abstract:
Environmental crisis and concerns about energy security are fueling the push for solar energy.
Organic photovoltaics (OPVs) offer an alternative for harnessing solar energy in feasible and
potentially low-cost manner. However, OPVs are suffering from low efficiency and instability
compared to conventional inorganic silicon solar cells. Fill factor and open circuit voltage are the
solar cell parameters that affect the solar cell efficiency directly. This thesis focuses on the
understanding of fill factor and open circuit voltage from the OPV perspective and their
correlation with other OPV parameters. In order to evaluate OPV parameters, inverted
architecture ITO/ZnO/P3HT: PCBM/MoO3/Ag OPV devices are fabricated and characterized
under dark and illumination. Various OPV parameters have been extracted from dark and light I-
V characteristics. Due to field dependent polaron pair dissociation in OPV, shunt and series
resistance model needs to be revisited. Simulation results of the OPV device show that the two
parasitic resistances are not able to describe fill factor completely. Two new fill-factor (FF) like
parameters, fill factor dark (FFD) and fill factor light (FFL), dependent on dark and photo-
generated current respectively, are used to explicitly reveal the impact of each current component
on fill factor. Correlation between fill factor, open circuit voltage, efficiency and Jd(Voc)/Jsc, are
also observed. Differential function G(V) of dark current characteristic is employed to estimate
the open circuit voltage of the OPV devices.
97
Title : A Study of Performance of Dual-Metal Gate Double-Gate
TFETs (DMG-DGTFETs) and Development of a new
Surface Potential based Drain Current Model
Author(s) : Prabhat Vishwa
Roll No : 13104159
Supervisor(s) : Dutta Aloke
Abstract:
A detailed study of the Dual-Metal Gate(DMG) Double-Gate Tunneling Field Effect Transistors
(DGTFETs) is presented here. Simultaneous variation of tunneling and auxiliary gate work
functions is shown to improve the current ratio between the ON and OFF states (ION/IOFF ratio)
and the average subthreshold slope. An optimum ratio for the tunneling to auxiliary gate length
that maximizes the device performance, is also extracted from simulations. We have
demonstrated that the device performance of DMG-DGTFETs with scaling does not degrade. A
comparison between the Dual-Material Gate (DMG) and Single-Material Gate (SMG) TFETs
shows that even for scaled devices, the DMG technique continues to give better performance
(higher ION/IOFF ratio and lower subthreshold slope) than the SMG technique. A 2D analytical
model for the surface potential of DMGTFETs is also presented here. It incorporates the effects
of drain and gate voltages, gate metal work function, insulator thickness, silicon film thickness,
and source and drain depletion. The band-to-band generation rate is integrated to compute the
tunneling current, considering the band-to-band tunneling to take place both at the source as well
as the drain depletion regions. The model accounts for variable drain doping through a fitting
function, which is based on the concept of generation current, and the results have been found to
predict the correct ambipolar behavior of TFETs. The results of our model both for the surface
potential and the tunneling current match very well with those obtained through TCAD
simulations.
98
Title : Development of a new Analytical Model for the Drain
Current of Double-Gate Tunnel FETs (DGTFETs)
Author(s) : Pal Arnab
Roll No : 13104020
Supervisor(s) : Dutta Aloke
Abstract:
In this work, we have developed an analytical model for the drain current of a Double-Gate
Tunnel Field-Effect Transistor (DGTFET), which is derived by using the pseudo-2D Poisson’s
equation in order to get the surface potential in both the channel and the source regions. We have
presented a new model for the surface potential, which is valid for the channel region being both
under depletion and accumulation, along with an analytical model of the phenomenon of source
depletion. We have also proposed a modi fication in the existing model for the band -to-band
tunneling current by taking into account the change in the carrier momentum vector during its
transit from the source valence band to the channel conduction band. Another important
contribution of this work is the prediction of zero drain current at zero drain bias, which existing
models failed to emulate, and is physically modeled forfirst the time here. The negative
conductance region shown by the transfer characteristics at low gate bias is also modeled for the
first time by using an empirical formulation, employing a solitary fitting function. Extensive
TCAD simulations were performed in order to prove the veracity of our model with respect to
drain bias, gate bias, substrate doping, and metal electrode, and the match was found to be
excellent.
99
Title : Small-Signal Modeling of Laterally Asymmetric MOSFETs
including Non-Quasi-Static Effects
Author(s) : Patel Naresh
Roll No : 10327435
Supervisor(s) : Chauhan Yogesh Singh&Dutta Aloke
Abstract:
The interest in accurate modeling of high voltage transistors like LDMOS and VDMOS has
increased in recent years due to the compatibility of these devices with standard CMOS
technology. However, the development of models for these high voltage devices depends on an
accurate modeling of the LAMOS, which is the building block of these devices. Since due to the
lateral non-uniform doping in the channel region, the Ward-Dutton charge partitioning scheme is
not applicable for LAMOS, hence, the admittance parameters of the LAMOS are determined by
solving for the small-signal real and imaginary components of current under the quasi-static
assumption, i.e., restricting the magnitude of the frequency. In this work, we develop the
admittance parameters without putting any restriction on the magnitude of the operating
frequency of the small-signal voltages. This thesis rst explores the small-signal modeling of
conventional MOS transistors under steady-state, quasi-static, and non-quasi-static situations.
The connection between these three situations is well elaborated, and it is illustrated that the
model under the non-quasi-static situation reverts back to the quasi-static model when the
operating frequency is low, and that in turn approaches the steady-state model, when the
operating frequency becomes very small or zero. Second, it covers the DC modeling of the
LAMOS, based on which, under the general non-quasi-static situation, the model for the small-
signal admittance parameters is developed. It is shown that at low frequencies, these parameters
show unusual capacitance characteristic exhibited by LAMOS. It is also shown that when the
doping gradient tends to zero, in the limit, the proposed model reverts back to the conventional
MOS transistor model. The results of our model for the y-parameters match very well with
TCAD simulations up to a frequency of 20 GHz
100
Title : Design and simulation of the thin film solar cell Copper
Indium Gallium Selenide (CIGS) and Copper Zinc Tin
Sulfide (CZTS) using ATLAS , SILVACO
Author(s) : Choudhary Nisha
Roll No : 10327453
Abstract:
The characteristics of copper chalcopyrite (CuInGaS_2 and Cu_2 ZnSnS_4) based thin film
solar-cells have been studied using ATLAS a device simulation framework from SILVACO. In
this study the simulated solar-cell characteristics obtained by varying the parameters like
thickness of absorber and window layers, doping, band gap, and grain-boundaries have been
correlated with the corresponding effects these parameters have on the photo-generation and
transport of electrons and holes across the device. The simulation results have also been
compared with the corresponding results available in the literature.
101
Title : Electrothermal Numerical Modeling of heterogeneous
binary transition metal oxide based RRAM
Author(s) : Asapu Shiva
Roll No : 10327152
Supervisor(s) : Chauhan Yogesh Singh
Abstract:
Resistive-switching random access memory (RRAM) has received a lot of attention for the
application as next generation non-volatile memory. Among the various materials investigated
over the years, transition-metal oxides have shown great promises due to their fast switching,
compatibility with the CMOS process and easy scalability. The fundamental challenge is the lack
of a proper theory that accurately describes the resistive switching phenomenon. However, a
conceivable theory has been developed over the years with the help of which modeling was
made possible. Modeling a single layer resistive switching cell has been done in the literature.
Heterogeneous bi-layers have been proposed recently in order to have a substantial control over
the RESET current in resistive switching devices. In this thesis, we model a heterogeneous bi-
layered stack of WOx/TaOx resistive-RAM (RRAM) or resistive switching (RS) cell. We use
finite element method (FEM) to solve a set of 3 self-consistent partial differential equations
(PDEs) using a numerical solver, COMSOL multiphysics. With two entirely different oxides
forming a stack, one of them (WOx) would act as the reservoir of defects due to oxygen
vacancies (Vo) and the other layer (TaOx) is the active layer wherein a conduction path
(conduction filament) would be established
102
Title : Modelling and Simulation of Laterally Asymmetric Channel
MOSFETS
Author(s) : Randhi Sai Kishore
Roll No : 13104116
Supervisor(s) : Yogesh Singh Chauhan
Abstract:
In this era of automation, requirement of High-voltage MOSFETs has increased manifold. The
integration of High-Voltage devices and advanced Low-Voltage CMOS devices is becoming an
increasing requirement for input/output interface of various System-On-Chip applications, such
as Sensors, Power Management, Flat panel Displays and RF Power amplifiers. In the first part of
the work done in this thesis, an Extended drain MOSFET (EDMOS) device is generated and
simulated by using Sentaurus Technology Computer Aided Design (TCAD) and the device
characteristics are observed. In addition we consider Dual Work Function Gate concept and
Super Junction concept to improve the device characteristics. In the second part of the work done
in this thesis, the spatial composition grading of a binary metal alloy is considered as gate
material for EDMOS transistor. Titanium with work function of 4.4 eV and Platinum (Pt) with
work function of 5.3 eV are considered elements for this binary metal alloy. The variation of the
surface potential for different channel lengths is calculated, and then is used to calculate the
threshold voltage for these devices. The threshold voltage obtained is then used in the threshold
voltage based compact model to get the I 󲐀V characteristics of the given devices. The results
obtained for these devices are compared with the data obtained from TCAD simulations for
different lengths of the channel. The fitting results of the model with the TCAD simulations
show quite good fitting.
103
Title : Investigation of Band to Band Tunnelling in Quantum-Well
MOSFETs
Author(s) : Kumar Rohit
Roll No : 10327610
Abstract:
This thesis attempts to understand and calibrate the Band to Band Tunneling (BTBT) in
Quantum Well (QW) III-V Metal Oxide Semiconductor Field Effect Transistor (MOSFET). This
is crucial for proper prediction of the lowest OFF current (IOFF) achievable in the low mass
(hence attractive), but concomitantly low band gap (concern with BTBT) materials. Here
In0.7Ga0.3As MOSFET of different channel and gate length will be investigated. We will focus
on experimentally varied device design parameters like drain –gate overlap, oxide thickness,
drain voltage, channel and gate length and temperature. The primary phenomenon that will be
studied due to the BTBT is the so-called Gate Induced Drain Leakage (GIDL). The main focus
of this thesis is to analyse the BTBT pattern in different device configurations, and to calibrate
the model parameters to match the experimental data. Proposals to reduce the BTBT leakage will
be proposed. In this thesis, simulation has been started with basic device structure and finally
experimental structure has been simulated.
104
Title : Dual and Triple Band Bandpass Filter realization Using
Split Ring and Spiral Resonator for Significant Size
Reduction
Author(s) : Kaur Priyajeet
Roll No : 13104106
Supervisor(s) : Biswas Animesh
Abstract:
In the era of wireless communication, various research fields are emerging. Microwaves are
providing excellent source to operate various communication devices. Researchers in the field of
Microwaves and RF are facing challenges at different levels of application developments. Filters
are fulfilling the requirements to segregate the required frequency bands, It covers an important
circuit in RF Transceiver. There has been an extensive research on this topic, Still new methods
are being designed so as to make designs as robust and efficient as possible. Planar filters have
an extra advantage of being extremely space efficient and thus can be accommodated on chip
along with other front end elements. The thesis presents design of Dual and Triple band
Bandpass filter using two different unit resonator cell. Split Ring and Spiral resonator properties
are utilized to design dual and triple band respectively. Various methods are utilized to enhance
the response of filter in Dual band filter design. Coupling properties are studied. Coupling matrix
approach is utilized to design Triple band filter design. Simulations are done using HFSS and
ADS Momentum. The Designed structures are fabricated in PCB Lab. Measurements are done
using Vector Network analyzer for the fabricated Designs. Results are analyzed and explained.
105
Title : Studies on Multilayer Substrate Integrated Waveguide Slot
Coupler and its Application to Microwave Circuits
Author(s) : Biswas Animesh
Roll No : 13104119
Abstract:
A resonating slot possesses the property to radiate as well as can be used to couple power from
one waveguide to another waveguide. In this thesis, Characterization of slot coupler has been
taken as the basic building block for designing of 1 to 2n way power divider. This thesis also
investigates substrate integrated waveguide (SIW) technique based slot couplers and its
application as antenna array feeding structure. Substrate Integrated Waveguide (SIW) is a
technique to design planer waveguide in which metallic via places in such a way so that it can
propagate power with minimum power leakage from the walls. SIWs exhibits compact size, low
cost and highly immune to the electromagnetic interferences in microwave and millimeter wave
technology. Behavior of coupling slots for a range of tilt angle, frequencies and resonant length
are presented. The proposed design principle has been described on the characterization results
of the double layered cross coupled SIW slot coupler. An application of the proposed SIW
coupler as 1 to 2 power divider is also presented. The multilayer behavior and SIW technique
make the structure compact which occupies less space as compared to traditional power dividers.
The proposed design works for 474 MHz (4.65% for -15 dB) of the bandwidth and can be used
as a feeding network for the SIW based waveguide slot array antenna.
106
Title : Studies on internal and external cloaks for improved
cloaking performance using realizable material parameters
Author(s) : Vura Pravallika
Roll No : 13104104
Supervisor(s) : Srivastava Kumar Vaibhav
Abstract:
Research on metamaterials has been growing ever since the first experimental realization of a double negative
medium. Metamaterials give enormous choice of material parameters for electromagnetic applications. The theory
of transformation optics(TO) provides a great tool to exploit the vast possibilities of the constitutive parameters of
metamaterials. The main tool associated with TO is that of coordinate transformations, in which isotropic space is
conceptually warped or otherwise distorted as a means of guiding the trajectories of waves. A lot of fascinating
optical devices that uses transformation optics have been identifi ed, especially invisibility cloaks. The aim of this
thesis is to adopt the techniques of transformation optics in order to design novel invisible cloaking devices that
contributes to improvement in cloaking performance and better realizability of material parameters. Firstly, a
parabolic transformation function instead of linear transformation function has been utilized in internal cloaking for
smaller dynamic range of material parameters with lower scattering. Two kinds of simplifi ed material properties
namely, reduced parameters and modifi ed reduced parameters are obtained by using the proposed parabolic
transformation function. An extra degree of freedom with parameter `p' is present in designed cloaks whose value is
governed by monotonicity condition. This parameter provides flexibility between impedance matching and dynamic
range of material properties. Both simplifi ed cloak designs have finite material property at the inner boundary. In
the case of modi fied reduced parameters, the cloak is having perfect impedance match at the outer boundary of the
cloak for any thickness of the cloak whereas impedance matching in the case of reduced parameters is dependent on
`p'. The scattering characteristics of the proposed quadratic cloak is compared with the linear cloak, signifi cant
improvement in invisibility performance is noticed for the proposed one with respect to the earlier reported linear
cloak. When a thin cloak shell is desired, the proposed cloak with modifi ed reduced parameters (that are realizable)
is much more superior than the linear cloak, as the cloaks outer surface is impedance-matched to free space for any
thickness of the cloak. Secondly, two composite shaped external cloaks are proposed based on the concept of linear
transformation method and complementary media. Each composite shaped cloak is sub-divided into regions of
rectangular and triangular shapes. The advantage of the proposed cloak designs is that the material properties are
finite, homogeneous and nonsingular within all sub-regions of the cloaks which is good for practical realization. The
cloak size of each of the proposed designs is smaller compared to the conventional external cloaks. Moreover the
cloaking area of proposed cloaks can be increased in the vertical direction without any change in material
parameters of core and complementary regions which is an additional bene fit for hiding long objects. Full-wave
simulation results demonstrate that the proposed complementary invisibility cloaks has a good performance for
cloaking multiple target objects Finally, to increase the cloaked area in the above design, we can interchange the
core & cloaked regions of above cloak design and introduce an identical structure symmetrical to it which gives
hexagonal shaped reciprocal cloak which possess all the advantages of the above design along with an additional
advantage of increased cloaking area. By utilizing the concept of transformation optics and complementary media, a
hexagonal shaped reciprocal external cloak is proposed. The hexagonal shape has been sub-divided into regions of
rectangular and triangular shapes. The rectangular sub-regions have diagonally anisotropic material properties
whereas the triangular sub-regions have complete anisotropic tensor properties. The advantage of proposed cloak is
that the material properties are homogeneous within all sub-regions. The other advantage is that its cloaked area can
be increased in X-direction without any change in material parameters. Fullwave simulations reveal that the
proposed cloak has good performance even for hiding multiple objects.
107
Title : Design of X-Band Power Amplifier based on GaAs-pHEMT
Technology
Author(s) : Makwana Milan Kiritbhai
Roll No : 13104073
Abstract:
The field of communication has seen significant and unprecedented progress in the past decade.
Power amplifiers are one of indispensable components in radar, modern communication systems,
and local multipoint distribution system (LMDS). The need for higher data rates, reduced cost,
and submicron size device technology makes MMIC power amplifier technologies at X-band
very attractive. As during transmission of signals through medium there are several losses occurs
so, it is required to amplify the signal with large power. This thesis presents a design of an X-
band power amplifier based on GaAs pHEMT technology. The design accuracy depends upon
the device model. It is imperative to have a meticulous device model for the broad frequency
range to use in simulation tools. For this measurement based millimeter wave bias dependent
model based on S-parameters for the 0.1 µm gate length GaAs pHEMT having gate peripheries
of 2x25 µm (number of gate finger x unit gate width) is fabricated on the 2mil (50.8 µm)
GaAs/AlGaAs/InGaAs substrate using PP10-10 process from WIN foundry. Measurement
includes pulsed I-V and on wafer scattering parameters for the frequency range of 45 MHz to
110 GHz for this device. In this work, a two stage power amplifier design across 9-11 GHz, X-
band using circuit simulation in Advanced Design System (ADS) is designed. The device
technology used for this design is PP10-10 process from WIN foundry. The resultant circuit
typically provides saturated output power of 23.5 dBm, a gain of 29 dB, input and output return
loss greater than 13 dB and 8 dB respectively. The power added efficiency (PAE) for the present
design is 22%.
108
Title : Design of X-Band low phase-noise oscillator and
comparative analysis of the two topologies
Author(s) : Shrivastva Vivek
Roll No : 13104160
Abstract:
The need of higher bandwidth, higher performance, functionality, size, power consumption
requirements with advances in technology has always been a topic of interest. In Present medical
facilities like remote surgery, high precision detection devices like X-rays, MRI require an e
cient source. In addition to this, Navigation systems like GPS, GLONASS and IRNSS etc.,
modern digital systems, communication systems and various RADAR technologies cannot be
imagined without an oscillator. In all, oscillators are the key components for almost every system
in present world and a small amount of unpredictable performance in an oscillator used in these
systems leads to the failure of a good and e cient system. The main aim of this thesis is to
develop a comparison in the two widely used microwave oscillator design approaches viz.
feedback approach and negative resistance approach. First, a novel high selectivity band-pass
microstrip lter has been designed. This lter has been used to provide su cient Quality factor to the
oscillator circuits according to Leeson's model for phase noise. The major portion of the thesis
covers the design methodology to design a microwave oscillator using above mentioned two
approaches. The operational band for the design in this thesis is X-band. As a concluding remark,
the last chapter highlights comparison between X-band negative resistance oscillator and X-band
feedback oscillator.
109
Title : Design of power amplifier
Author(s) : Sharma Sonu
Roll No : 13104145
Abstract:
Power amplifier is an important key member in communication. Power amplifier as high output
power amplifier or high efficiency power amplifier is used at the transmitter side and low noise
amplifier is used at receiver side. The GaAs pHEMT is very good device to use as power
amplifier. This work is divided in two part, in the first part device is modelled so as to get same
response after the fabrication. Generally design tool gives the device which is almost different
from fabricated device because after the fabrication process it has many non-ideal components
which does not give the similar response to the simulated design. Thus the modelling has been
done for different bias points. In the other part I have designed the class A power Amplifier
which gives 13.250 dBm output power and its power gain is 8.917 dB at 15 GHz operating
frequency. The device is used for the design of power amplifier is GaAs pHEMT, whose gate
width is 25 µm and length is 0.1 µm. In the end I have shown some trade-offs in the design of
the power amplifier which are between efficiency and linearity, between power gain and
efficiency and between power gain and output power. Power added efficiency of the power
amplifier is 36.722%. Ideally efficiency of class A power amplifier should not be more than
50%.
110
Title : Design of Broad Band Bias-T on Alumina Substrate
Author(s) : Kumar Sonu
Roll No : 13104144
Supervisor(s) : Das Utpal
Abstract:
Abstract:
In this work, microstrip line based Bias-Ts have been designed using HFSS software to work in
the frequency band 4.8-100GHz on alumina substrate. In this design a taper has been introduced
in the high frequency (HF) line to improve the isolation of the DC line from the signal HF. A
novel MIM capacitor has been introduced in the HF line to obtain DC from one of the HF ports,
which did not impede HF transmission. Two alumina substrate thicknesses, namely 0.65mm and
0.1mm, have been used for the design. For the 0.65mm substrate, an HF transmission of ~ -4dB
from 4.8-50GHz has been obtained with a minimum isolation to the DC port of ~ -25dB or better
from 4.8-50GHz. The corresponding VSWR is 1.02-3.50 in the same HF range. The masks for
the fabrication have also been prepared and initial process steps performed. The 0.65mm alumina
substrate device showed limitations at ˃ 30GHz while Bias-T design on the 0.1mm substrate
shows an HF transmission > -3dB for 7.5-100GHz band except for 17.3 and 85.4GHz (-3.96dB
and -3.28dB). Isolation from DC port is < -20dB and the corresponding reflection is below -8dB
in the 7.5-100GHz band. The VSWR being 1.04-2.40 puts this device performance at par with
the industry standards. The first design can easily be implemented for Bias-T for 40Gbps
photodiodes, using V-connectors, which were purchased and the package designed, but could not
be completed. On the other hand, the second design, although similarly implementable, needs the
more expensive and more difficult to handle, W-connectors. Measurements facilities are also not
easily available in this frequency range of ~ 100GHz. However, it is left to a further researcher to
implement it which would find useful applications in the W-band (75-100GHz).
111
Title : Design of broadband Bias-T on GaAs substrate
Author(s) : Shah Mithil
Roll No : 13104136
Supervisor(s) : DasUtpal
Abstract:
Large bandwidth operation up to 50 and 100GHz Bias-Ts have been designed on GaAs
substrates of thicknesses 450um and 200um, respectively. The dimensions for the 100GHz Bias-
T is 9.52mmX10.42mmX0.2032mm. 350nm Au conductors is used for microstrip conductor and
ground plane design. For adhesion of Au on GaAs, a 50nm Ti layer is designed to be coated on
GaAs before Au coating. Both HFSS and ADS software has been utilized for the design. The
design uses three butterfly stubs for HF isolation to the dc bias path. The butterfly stubs are
centred at 3.75, 12.58, and 19.58GHz of angles 15°, 60°, and 45°, respectively. The stubs were
placed at approximately λg/4, λg/4, and 5λg/4 from the T-junction, respectively. The minimum
HF isolation to the dc path achieved is 20dB from 4.5-100GHz. To eliminate the detrimental
notches in S21 at frequencies between 10-20GHz and 30-40GHz, a 10-42Ω microstrip taper is
used in the design. The length of taper on either side of the T-junction is λg/4 at ~5.5GHz. A
novel inline capacitor comprising of two conductor separated by ZrO2 insulator has been added
on the taper at the high-frequency port side for dc blocking. This can be easily achieved on GaAs
using standard photolithography and lift-off technique. A flat transmission response (S21) is
obtained with a maximum of ~2.9 dB insertion loss at ~38 GHz. Isolation of >25dB is obtained
on the entire band of 4.9-100GHz except 16dB in a narrow band around 38 GHz. A return loss of
7dB is obtained upto 6GHz and >10dB is obtained upto 100GHz. An advantage of this design is
that using V-connectors and without the m thick GaAs substrate one can operate the device
toµneed to use a 200 50GHz. For 100GHz operation lapping of a standard 450um thick GaAs
would be necessary.
112
Title : Design of Planar Resonator Sensor for RF Testing of
Liquids
Author(s) : Samant Himanshu
Roll No : 13104053
Supervisor(s) : Akhtar M Jaleel
Abstract:
Measurement of dielectric properties of liquids in the RF and Microwave frequency range has got various
applications such as contamination detection in agricultural and food products, testing of blood samples
for biomedical applications, microwave heating in polymer and composite industry, etc. The
contamination in various food products can be detected by observing the changes in the dielectric
properties of the food sample as comparing to that of the reference sample. Similarly the change in
dielectric properties of infected blood samples with respect to these properties of the blood samples of a
healthy person can be monitored for some biomedical applications. In this work, a non invasive planar
resonator sensor is developed for determination of dielectric properties of liquid samples, which has then
been tested for detection of contamination in various reference samples. The designed sensor is made
non-contacting by covering the resonator area with a thin polypropylene sheet. The proposed resonator
sensor is first realised on the central metallic strip of a coplanar waveguide, whose characteristic is similar
to that of a band stop filter. The proposed resonator is compared with the interdigital capacitor (IDC)
resonator, which is also realised on a coplanar waveguide. The proposed resonator sensor is found to have
better sensitivity than the IDC resonator sensor and hence appears to be more appropriate choice for RF
testing of liquid samples. A numerical model of the proposed resonator sensor is developed using the
electromagnetic simulator, the CST studio, for determination of the complex permittivity in terms of the
resonant frequency and quality factor. Effect of the polypropylene sheet on the sensitivity of the sensor
has been analyzed. The designed coplanar resonator sensor is fabricated on a FR4 substrate, and a hollow
cylindrical container is placed around the resonator area to contain the liquid samples by taking its effect
into account. The loading of sensor with the dielectric test specimen changes the capacitance of the
proposed resonator which causes a shift in the resonant frequency. The perturbed frequency is used for
the measurement of the dielectric properties of liquid samples. The numerically established relations are
experimentally verified for various liquid samples like Benzene, Ethyl Acetate, Pentane, Ethanol, water
etc. It has been observed that the proposed coplanar resonator sensor can accurately characterize various
chemical solvents with a maximum error of 4 %. The proposed technique is then extended to design a
dual band coplanar resonator sensor which may be advantageous for the characterization of dispersive
liquids. The proposed dual band sensor is designed to operate at the two ISM (Industrial Scientific and
Medical) bands. The permittivity values of different liquids calculated using the proposed sensor are
compared with the data available in literature.At last, the proposed resonator is realised on a microstrip
line in order to compare its performance as compared to the CPW based sensors. An experimental
modelling is proposed using this resonator sensor to determine the molarity of acetic acid solution in
terms of fractional change in the resonant frequency of the resonator. The proposed microwave method to
determine the molarity of liquids can be advantageous under certain situations as compared to that of the
conventional titration method which is actually time consuming for some cases due to the involved
chemical reaction, and where the sample under test cannot be recycled.
113
Title : Design of Multi Slot Vivaldi Antenna with Improved
Directivity for Microwave Testing Applications
Author(s) : Kumar Pankaj
Roll No : 13104098
Abstract:
In this thesis, the multi slot Vivaldi antennas are designed, fabricated and tested. The Vivaldi
antenna is basically a travelling wave end fire antenna, which is well-known for ultra-wide band
(UWB) characteristics. These types of antennas are being used to for various applications such as
the pulse radar, remote sensing, military mobile wireless systems, microwave imaging etc. The
microwave imaging technique is presently being explored to be used in airport security, aerial
surveillance, through wall imaging etc. Currently, to achieve good spatial resolution and
penetration of EM-wave through the object under test, there is a huge interest to develop UWB
antennas with high gain and narrow 3dB beamwidth. The Vivaldi antenna having good
directivity and UWB characteristics appears to be a viable candidate for the microwave imaging
applications. However, the conventional Vivaldi antennas exhibit somewhat low directivity and
high 3dB beamwidth than the desired value although they possess quite good UWB
characteristics. It is mainly due to this reason that in this work, the modified Vivaldi antenna
having multi slots has been proposed to achieve higher directivity and narrower 3dB beamwidth
without compromising its UWB characteristics. In this framework, first a compact double slot
Vivaldi antenna (DSVA) for 2-10GHz frequency band with improved directivity ranging from
3.5-10.0 dBi is designed, fabricated and tested. Afterwards, an anisotropic zero index
metamaterial (AZIM) cell was loaded into the E-plane of the designed DSVA to improve its
directivity. The measured results of AZIM loaded DSVA are found to be in good agreement with
the simulated results. In the next step, the four slot Vivaldi antenna (FSVA) is proposed in order
to further improve the directivity and to reduce the 3dB beamwidth as compared to the DSVA. In
order to feed the four slot Vivaldi antenna, the Wilkinson power divider and the T-junction
power divider are proposed and compared on the basis of their performance in the operating
frequency band. It was found that the T-junction power divider using binomial transformer had
wider bandwidth. Also, unlike the Wilkinson power divider, the T-junction power divider
usually does not require any lumped element. Hence, the proposed T-junction power divider is
subsequently used to feed the newly proposed FSVA. The directivity of the designed FSVA is
found to improve typically by 4 dBi, while the 3dB beam width is typically reduced by 30° in the
operating frequency band as compared to that of the DSVA. The fabricated FSVA has been
tested for the microwave imaging and testing to reconstruct the image of one and two objects
hidden inside the test media.
114
Title : Continuously Tunable Multi-wavelength Ultra-Fast Fiber
Ring Laser
Author(s) : Chandra Nishanth
Roll No : 13104094
Supervisor(s) : Kumar Pradeep
Abstract:
Ultra-short pulses in mode-locked lasers are a topic of extensive research due to their wide range
of applications from optical clock technology to measurements of fundamental constants of
nature and ultra-high speed optical communications. Ultra-short pulses are especially important
for the next generation of ultra-high speed optical systems and networks operating at 100Gbps.
Pulse sequences with pulse-width on the order of few picoseconds are best probable choice for
the generation of multi-Gbps optical carrier data. In this thesis, a continuously tunable multi-
wavelength ultra-fast fiber ring laser at 10 GHz repetition rate is studied experimentally. A novel
way to induce polarization and wavelength dependent loss in the laser cavity is proposed. This
makes the mode-locked laser tunable over a wide range of C band and also makes multi-
wavelength operation possible. The stability of the laser is measured and a software based
feedback loop is developed to improve the laser’s long term stability. Finally, a method is
proposed to generate other useful pulse shapes like rectangular waveforms from the ultra-short
pulses obtained.
115
Title : An Inverse Scattering Approach To Design Small Size
Microstrip Filters Using Defected Ground Structures
Author(s) : Sannake Rohit
Roll No : 13104124
Abstract:
Planar microwave filters have gained much popularity in RF applications due to their light
weight, low cost, and easy integration to active devices. Conventionally many analytical and
numerical methods exist for microwave filter design. In recent years, the inverse scattering
approach has been explored to design RF and microwave filters. One of the advantages of the
filter design using the inverse scattering approach is that the structure geometry can directly be
obtained in terms of the reflection coefficients, and even continuously varying transmission lines
can be realized. The Riccati differential equation approach is one of the inverse scattering
methods, which provides a stable solution for filter design as compared to other inverse
scattering approaches. However, one of the bottlenecks in the design of RF filters using the
inverse scattering approach has been the large size of the filter in order to realize the specified
response. In this thesis, a new method is proposed for length minimization of the filter which is
designed using the inverse scattering approach. In this technique, the effective length of the
designed microstrip filter is increased using the Defected Ground Structures (DGS). This
basically means that the characteristics of the filter remain almost unchanged even after reducing
the actual physical length of the designed structure. The periodic DGS increases the effective
inductance and capacitance of the microstrip line, which causes slow wave effect in the device.
Hence, the actual effective length of the filter increases through this DGS for the same physical
length. Different types of DGS are employed here to reduce the overall size as well as to improve
the performance of the filter. Based on the proposed approach, low pass and band stop microstrip
filters are designed, and their performance is first verified using independent simulation tools.
After validating the design, these filters are fabricated on FR4 substrate, and their characteristics
are measured using the network analyzer in the designated frequency bands. The proposed
method provides an overall reduction of the length of low pass and band stop filters by almost
60% and 30 % of their original lengths, respectively.
116
Title : Fractal Geometry Inspired Polarization Insensitive
Microwave Absorbers for Multiband and Broadband
Applications
Author(s) : Praneeth Munaga
Roll No : 13104084
Supervisor(s) : Srivastava Kumar Vaibhav
Abstract:
With the advent of radar, the necessity of passive devices that are capable of absorbing these
radio waves has also emerged resulting in various designs of microwave absorbers. These
microwave absorbers are designed in a number of frequency domains ranging from microwave
to infra red regime. Some of the important applications of the absorbers are solar cell, bolometer,
anechoic chambers, stealth technology, reduction of radar cross section etc. Because of massive
and bulky nature of the conventional microwave absorbers they require large area for
installations and cannot be used in practical applications. The solution of achieving near unity
absorption with light weight and very thin structures can be facilitated by making use of unusual
electromagnetic properties of metamaterials. However the absorption of these metamaterial
absorbers is limited to a narrow band of frequencies thereby making them not suitable for
wideband and multi-band applications. Hence many design methodologies have been proposed
for multi-band as well as broadband applications with significant increase in lattice dimension.
This thesis focuses on miniaturizing metamaterial absorbers for multi-band and broadband
applications. The compactness of the proposed metamaterial absorbers is retained by making use
of the space filling characteristic of the fractal geometry. Firstly, a dual-band polarization-
insensitive, wide angle metamaterial absorber has been designed comprising of tetra arrow cave
shaped unit cells. The compactness of this absorber structure is compared with other reported
structures. Secondly, a concentric Minkowski fractal loop based absorber structure has been
proposed which is capable of absorbing six distinct frequencies. Then by tuning the geometric
dimensions of this absorber structure, bandwidth enhancement at two different frequency bands
(X-band and Ku-band) has been achieved. Finally, a broadband absorber based on mounting of
lumped resistor method has been designed for C-band applications. The unit cell of this absorber
structure comprises of inverted Minkowski fractal loop along with lumped resistors. The
compactness of this structure is also compared with earlier reported structures.
117
Title : Analysis and Design of Ultra Wideband Wearable Antennas
Author(s) : Yadav Rajesh Anand
Roll No : 13104114
Supervisor(s) : Harish A R
Abstract:
Ultra Wide Band technology promises many innovative and commercially feasible applications.
It is expected by many visionaries that these ultra wide band technology enabled Devices can be
employed in body wearable and portable consumer products. Antenna for wearable and portable
applications have to face severe size and volume constraints due to limited available size and
mounting issues. Therefore antenna having compact size and small form factor are highly
desirable for commercial applications. High data rate applications demands that the radiation
pattern behaviour remains consistent across the bandwidth along with impedance performance.
These difficulties and challenges faced by designer leads to differentiation of these UWB
antenna designing from conventionally available antennas having narrowband characteristics. In
this thesis work three low profile antennas are fabricated on two different substrates (leather and
FR4).The measured results are compared with simulated results in terms of reflection coefficient
and radiation patterns. For these antennas to work properly, the ground plane size is required to
be electrically much larger so as to approximate an infinite ground plane and thus occupies a
larger portion of the overall system. For designing a compact antenna ground plane size need to
be minimized. Generally antenna structures are simulated in EM simulators without considering
feeding cable so as to reduce computational burden on it. However, when final product is
prototyped for measurements it has to be fed through feeding cable. Due to small physical size of
antenna ground plane the currents flow back on outer surface of cable. The standing wave pattern
is formed on the feeding cable and will result into radiation which disturbs original radiation
pattern. In order to minimize the ground plane effects, L shaped slots are introduced in ground
plane. Thus the current distribution in ground plane can be altered to reduce effects of small
ground plane size. When slot is not introduced in ground plane current distribution is a function
of frequency. With introduction of slot on the ground current distribution is confined by the slots.
Even at higher frequencies current distribution is slightly disturbed still it is confined by the
slots. This leads to less dependency of antenna performance in terms of ground plane size. In
order to understand the working of antenna or predicting its behaviour when integrated with
system and used in communication system equivalent circuit model was derived for a better
understanding of antenna working principle. The Vector fitting algorithm was used to derive
equivalent model for antenna structure. Measurement of electrical parameters of any material
helps in modelling that material in simulation environment which leads to minimum difference
between simulated results and measured results. Characterising materials before using them in
antenna design is the first step in antenna design workflow. The leather is characterised in this
thesis with sole purpose of integrating antenna in body worn applications. To study the effect of
human body on the performance of antenna, it is placed on three layered human model.
118
Title : Compact And Enhanced Bandwidth Polarization
Reconfigurable E-Shaped Patch Antenna
Author(s) : Velma Reddy Chandrashekar Reddy
Roll No : 13104154
Supervisor(s) : Harish A R
Abstract:
In modern communication systems, the antenna which can perform different tasks depending up
on the environment is more desirable. Generally a reconfigurable antenna refers to that antenna
which has the capability of changing polarization, radiation pattern or frequency of operation.
Their diversity and agile nature created a new dimension of applications in Multiple Input
Multiple Output systems, Cognitive radio, satellite and many other systems where
communication is involved. Reconfigurable antennas have increased functionality such as
producing different polarizations, beam steering etc., in the confined volume. The aim of this
thesis is to design a compact and enhanced bandwidth polarization reconfigurable E- shaped
patch antenna. The compactness in the antenna can be obtained by removing the λ/4 DC bias
stub which occupies 20% of length and by introducing a C-slot inside the outer patch of the E-
shaped patch antenna reduces the size of the antenna. Introducing shorting pin at the appropriate
positions in the antenna between the upper patch and the ground provides DC return path. By
changing the feeding technique, not only the individual bandwidths of axial ratio (25%) and
input reflection coefficient (19.23%) are improved but also the overlap of frequency bands of
S11 (< -10 dB) and axial ratio (< 3 dB) has improved well from 7% to 19.23% with a 5.5 dBic
maximum realized gain. Practical issues like availability of pin diodes and implementation of
shorting pins are considered and the shown results are discussed to validate the new changes
incorporated in it.
119
Title : Design, Fabrication and Characterization of a Vacuum
Gated Diode
Author(s) : Vashistha Sachin
Roll No : 13104128
Supervisor(s) : Das Utpal
Abstract:
High - speed electronic devices rely on short carrier transport times, which are usually achieved
by decreasing the channel length and/or increasing the carrier velocity. Scaling down of the
semiconductor transistor has reached its physical limit and also their speed of operation. Vacuum
is the superior carrier conveying medium than any other solid materials, because electrons do not
suffer from scattering with the atoms as in semiconductors. Thus limiting speed of operation of
the device is determined by RC time constant only. This type of vacuum channel transistor is
also useful in display devices mostly made on glass or a dielectric sheet. A detailed study of a
vacuum microelectronic devices is presented here. Two types of vacuum gated diode have been
analyzed, namely (a) metal electrodes and (b) InGaAs electrodes. Channel lengths considered is
20nm. A simulated operational voltage of 10 – 35V shows current flows were obtained in the
range of 50 - 1000μA. A current variation of 50 - 500μA was obtained for a gate voltage
variations of 10 – 35V for a fixed anode – cathode bias of 10V. For the highest currents a
maximum power dissipation for the device is found to be of ~1nW. Fabrication steps for the
diode and triode devices using photolithography and FIB is also presented here. Result of a
fabricated Al-emitter diode shows that the anode-cathode current measured is almost one third of
the simulated value. This discrepancy is attributed to non-uniform anode as well as anode edges
arising from uncontrolled fluctuations in the focused ion beam used for milling the nano-scale
structures. At 35V the device burnt at a current of ≤500µA due to local heating at the imperfect
edges. Hence the triode fabrication could not be implemented.
120
Title : Design implementation and control of low voltage DC
microgrid
Author(s) : Munagala Chaitanya Sai
Roll No : 12104043
Supervisor(s) : Das Shyama Prasad
Abstract:
In the present work, a basic DC microgrid comprising of utility grid connecting unit, PV
generation unit and battery storage unit, is designed, implemented and controlled. A voltage
source inverter for grid connection, a bidirectional converter for battery interface and a simple
boost converter for PV interface have been used. These three units are implemented and
controlled independently. The Utility Interface Converter (UIC) is modelled in synchronous dq
reference frame under both balanced and unbalanced utility grid conditions. Voltage oriented
control scheme in dq reference frame is used for rectifier current control. In general, microgrids
are connected to utility at distribution level, which is prone to get unbalanced due to single phase
loads. When the rectifier is connected to an unbalanced utility, second harmonic oscillations are
generated in injected power, due to interaction between opposite sequence voltages and currents.
An improved control strategy is proposed to attenuate these oscillations, while interfacing
microgrid to unbalanced utility grid. The PV Interface Converter (PVIC) and Battery Interface
Converter (BIC) are modelled in state space averaging approach. To emulate the behaviour of
renewable source, PV generation unit is operated in constant power mode. The BIC charges the
battery in buck mode and operates in boost mode while discharging. For both charging and
discharging regimes, control strategies are designed and implemented. A distributed control
strategy based on terminal parameters of the units is proposed to control the DC microgrid. To
suppress the circulating currents between units when connected in parallel, droop method is
used. Simulation analysis is carried out in MATLAB SIMULINK to evaluate the performance of
proposed control schemes. A Lab prototype type has been developed for experimental analysis.
Controllers are implemented in TMS320F28335 DSP experimental kit. Hardware results show
good correlation with theoretical and simulation analysis.
121
Title : Hall Effect Current Sensor using Transconductance
Amplifier
Author(s) : Sudheer Jillelamoodi
Roll No : 12104027
Supervisor(s) : Sensarma Partha Sarathi
Abstract:
All the modern conveniences today rely heavily on the current sensors. Rail current is a very
important to be measured in traction environment, the need arose for a reliable, large bandwidth
and dynamic range rail current sensor. A current sensor with high bandwidth and good dynamic
response is desired in these applications. For that application, 1A accuracy for the DC
component measurement is sufficient and is performed by a Closed loop Hall effect transducer.
Closed loop Hall effect based transducer also have fast response time, support high frequency
signals of the current from the high switching frequency of the inverter, and to achieve accurate
control of the speed. In the present work a closed loop Hall effect current sensor is designed and
implemented. Initially a closed loop Hall effect current sensor is modeled and the model is used
in designing the compensator. The transconductance amplifier is constructed using linear
MOSFETs. All the issues in the design process and limitations are addressed. The experimental
results of the transconductance amplifier as well as overall closed loop circuits are also
discussed.
122
Title : Observability analysis and topology error processing of
power systems in the presence of hybrid measurements
Author(s) : Sharma Neeraj Kumar
Roll No : 13104088
Supervisor(s) : Chakrabarti Saikat
Abstract:
Phasor measurement units (PMUs) are increasingly being used in modern power systems due to
their high reporting rate and accuracy. The majority of the measurement devices in the existing
power systems, however, are conventional asynchronous measurements. State-of-the-art state
estimation algorithms are therefore required to include the hybrid set of measurements.
Observability of the system with the given set of measurements is a pre-requisite of any state
estimator. This thesis proposes a numerical algorithm for observability analysis of power
systems having hybrid measurements. The proposed method is demonstrated with the help of a
number of test systems. Topology processing is a vital function of a power system state
estimator. Detection, identification, and processing of the topology error are pre-requisites for
successful execution of the state estimator. This thesis proposes an optimization technique for
topology error detection of power systems in presence of hybrid measurements. The proposed
technique is tested on the number of test systems.
123
Title : Optimal demand response for PV-Integrated households
Author(s) : Naik Kishankumar M
Roll No : 13104068
Abstract:
Electricity is the principle driving force of the modern world. The demand for electricity is
higher than ever. The stress on the power system is increasing continuously, with increasing peak
demand. The increase of conventional generation units such as power plants cannot keep up with
the ever rising demand. There is need for better demand management, greater local generation,
and uninterrupted power supply. Demand Response programs are the key drivers of demand side
management. In this thesis, a demand response algorithm is proposed to optimally schedule the
residential appliances, while incorporating generation from the photovoltaics (PV). An
optimization problem is defined to minimize the cost incurred by the residential customer.
Among the constraints, customer load preference, appliance ratings, and solar PV generation are
considered. This optimization problem is then solved in MATLAB to obtain a scheduling vector
for each controllable appliance. With the obtained optimal schedule, peak shaving/shifting is
achieved and the customer is able to save significantly on the electricity bill. Rooftop solar PV
produce clean energy locally and is a viable option for household applications. With the use of
battery, the issue of interruption in power supply is addressed. A grid connected solar PV system
and a backup battery system, connected to a IEEE 13-node test feeder are simulated in Real
Time Digital Simulator (RTDS). The e ectiveness of the proposed demand response scheme with
local generation and uninterrupted supply is verified.
124
Title : Online Health Monitoring of Aluminium Electrolytic
Capacitors For Single Phase Grid Connected PV System
Author(s) : Arya Abhinav
Roll No : 13104003
Supervisor(s) : Anand Sandeep
Abstract:
DC-link aluminum electrolytic capacitors (AECs) are used in single phase solar PV inverters to
suppress second harmonic and switching voltage ripples. Electrolytic capacitors are less reliable
than other components of PV system. Degradation of these capacitors affects the performance of
the system. Reduction in capacitance and increase in equivalent series resistance (ESR) due to
ageing increases ripples in dc voltage. DC-link voltage ripple causes oscillations in PV operating
point thereby reducing PV power extraction efficiency. The worsen effect of growing dc-link
voltage ripples is the breakdown of complete system. Two different techniques are proposed in
this thesis for online health monitoring of AEC. First technique is based on the online monitoring
of PV power extraction efficiency. Electrolytic capacitor is selected for power extraction
efficiency greater than 99%. Replacement of capacitor is suggested based on deterioration in
extraction efficiency. However, frequent replacement of capacitor increases cost of maintenance.
Therefore, optimal value of extraction efficiency is determined in this paper, at which total
payback period including maintenance cost is minimized. Second technique is based on the
extraction of capacitor impedance at twice the grid frequency. The capacitor current is estimated
from inductor current and the states of the switches. This eliminates the need of a large
bandwidth current sensor for capacitor current. A narrow bandpass filter is used to extract the
second harmonic component of voltage and current of capacitor. The ratio of the two gives the
second harmonic impedance. Suitable limit on the impedance is defined, after which capacitor
should be replaced. Detailed simulation studies are carried out using Simulink and results are
included. Both the proposed techniques are evaluated by experimentation on a scaled-down
laboratory prototype of solar inverter, developed for the purpose.
125
Title : An Optimal Smoother Based Dynamic State Estimator for
Power Systems Considering PMU Measurements
Author(s) : Sreenath J G
Roll No : 13104056
Abstract:
State estimation of the electric power transmission network has become an integral part of any
modern control centre around the world and it provides the best estimates of the system states.
State estimators ensure the secure operation of a power system. The development of
synchronized measurement technology (SMT) has opened new avenues for the dynamic
monitoring of the system states. This study introduces a power system state estimator based on a
two pass algorithm known as the Rauch-Tung-Striebel (RTS) smoother. The RTS smoother is a
two pass algorithm consisting of a standard Extended Kalman Filter (EKF) based forward pass
and a backward recursion smoother. The state estimator incorporates conventional Supervisory
Control and Data Acquisition (SCADA), as well as Phasor Measurement Unit (PMU)
measurements and the system states are estimated when measurements are obtained from
SCADA and/or PMU. The complex current measurements obtained from the PMU are directly
incorporated into the state estimator. A dynamic model of the system is developed considering
the system to be in a quasi-static state. The unknown parameters of the power system dynamic
model are identified using Holt’s 2- parameter linear exponential smoothing technique. The
results show that the RTS smoother can improve the estimation accuracy. The IEEE 14, 30, and
118-bus systems are taken as the test bed for the study. The proposed DSE is validated under the
real-time environment using Real Time Digital Simulator (RTDS), a real-time simulation tool, to
assess its effectiveness for online visualization of the power system.
126
Title : Tunnel FETs and Its Application to Digital Circuits
Author(s) : Kumar Deepak
Roll No : 13104037
Supervisor(s) : Mishra Santanu Kumar
Abstract:
In advanced technology, the down-scaling of conventional MOSFETs has led to an impending

power crisis, in which static power consumption is becoming too high. In order to improve the
energy-efficiency of electronic circuits, small swing switches are interesting candidates to
replace or complement the MOSFETs used today. Tunnel FETs, which are gated P-I-N diodes
whose on-current arises from band-to-band tunneling between source and channel. These new
devices can be attractive for low-power applications due to their low off-current and their
potential for a small sub-threshold swing. This thesis studies some TFET based digital circuits
and their performances. Since analytical models for Tunnel FETs are not available so the best
way for circuit simulations is to build a Look up table based model using Verilog-A. Both the N
and P type Tunnel FETs devices are simulated in Sentaurus 2D to generated look up table. This
model is used to simulate static and dynamic behavior of Tunnel FET based digital circuits in
cadence spectre.
127
Title : Power Sharing in Low Voltage DC Microgrid Using Low
Speed Communication
Author(s) : Agarwal Shyam
Roll No : 13104140
Abstract:
Renewable energy resources are gaining popularity due to depleting fossil fuel and increasing
energy demand. Microgrid allows efficient and reliable utilization of these renewable sources in
both grid connected and standalone operations. Due to increasing use of dc sources and loads,
low voltage dc microgrid is gaining popularity. To control the operation of dc microgrid
centralize schemes are suggested in literature, which uses central controller with high speed
communication. Presence of central controller limits the reliability of the system. Voltage-
current droop controller based distributed control scheme is used for power sharing between
sources. However, due to interconnecting cable impedance, trade-o_ exists between power
sharing and voltage regulation. To address this limitation a dynamic droop gain adjustment
technique is proposed in this thesis. This technique uses decentralized controller with low
bandwidth communication to achieve equal power sharing and good voltage regulation. The
source current of each converter is measured at low sampling rate and communicated to all other
converters using Controller Area Network (CAN) communication technique. The communicated
values of converter currents are used to adjust the droop gain in a decentralized manner, thereby
improving the power sharing among converters. Key advantages of the proposed technique are
equal power sharing, low voltage regulation and high reliability. Detailed small signal model of
the proposed controller is developed. Delay due to sampling rate and communication is modeled.
The stability of the system is analyzed by determining the eigenvalues. Sensitivity of eigenvalues
to communication delay is observed with the help of root locus plots. Performance of the
proposed method is verified by detailed simulation and experimentation studies. Eigenvalues to
communication delay is observed with the help of root locus plot.
128
Title : Zero Net Energy House Cluster: Protection and Energy
Management
Author(s) : Pandey Gaurav
Roll No : 13104047
Supervisor(s) : Singh Sri Niwas
Abstract:
The main objective of the present work is to create a smart dc micro-grid capable of 100%
autonomous zero net electric energy in the cluster of buildings to facilitate a low-carbon
sustainable electricity supply system under considerations of urban and rural scenarios. The
propose model comprises of house clusters with an autonomous communication developed for
the residential area, taking the consideration of stochastic behaviour of wind and solar
irradiation. Voltage droop and Slope compensation peak current mode control technique for the
bidirectional converters is employed for the Energy Storage System (ESSs). The bidirectional
converter stage for house clusters plays a pivotal role in standalone operation, in case a battery
pack is laid off from any house cluster, the dc bus voltage still be stabilized due to the proximity
bidirectional converter stages of other house clusters or community bank. The houses in the
cluster comprises of Permanent Magnet Synchronous Generator (PMSG), Solar Photovoltaic
(SPV), battery bank and variable load. The dc protection schemes broadly classified as the unit
and non-unit protection schemes under fault conditions is also investigated. The proposed model
is simulated on MATLAB/ Simulink environment.
129
Title : Development of Control Strategies for Power Management
with Renewable Resources
Author(s) : Jha Rahul Ranjan
Roll No : 13104111
Supervisor(s) : Srivastava S C
Abstract:
Renewable Energy Sources (RESs) are being increasingly deployed in the electrical networks, to
address the growing environmental concerns and to ensure energy security. Amongst various
types of renewable sources, Solar Photo Voltaic (SPV) and the Wind Turbine (WT) plants are
the most promising. Output of these sources is intermittent in nature and, hence, energy storage
devices are required to store surplus energy and feed to the load when needed. Proper operation
of these sources to ensure voltage control and power management requires use of suitable
converters. For the integration of the SPV to the utility grid, 3-phase Voltage Source Converter
(VSC) has been used to convert DC into AC. Three types of controllers, viz. PID, Fuzzy Logic
Control (FLC) and Adaptive Network based Fuzzy Inference System (ANFIS) have been
implemented for the DC link voltage control and the VSC current control. The model of the
Doubly Fed Induction Generator (DFIG), used for the wind plant, in the dq reference frame, and
the control of the rotor side as well as the stator side converters are developed in this thesis.
Battery Energy Storage System (BESS) is used, and integrated with the SPV and the wind
turbine, connected to the utility grid. A bidirectional three phase VSC is used for charging and
discharging of the battery. To develop controls for the home PV systems, a cluster of four houses
has been considered, in which each house comprises of a SPV and a battery. The power
requirement of the load is met either by the SPV or the battery. The power management among
the houses is ensured by the proposed controllers in the islanded mode.
130
Title : Object Matching Based on Speeded Up Robust Features
Author(s) : Annepu Y V Harsha Vardhan
Roll No : Y9329105
Supervisor(s) : Verma Nishchal Kumar
Abstract:
Autonomous object counting system is of great use in retail stores, industries and also in research
processes. A Speeded Up Robust Feature (SURF) based robust algorithm for identifying,
counting and locating all instances of a defined object in any image, has been proposed. The
defined object is referred to as prototype and the image in which one wishes to count the
prototype is referred to as scene image. The algorithm starts by detecting the interest points for
SURF in both, prototype and scene images. The SURF points on prototype are first clustered
using density based clustering; then SURF points in each cluster are matched with those in scene
image. The SURF points in scene image that have been matched w.r.t. a single cluster, are
clustered using the same clustering algorithm. Each cluster formed in scene image represents an
instance of prototype object in the image. Homography transforms are further used to give exact
location and span of each prototype object in the scene image. Once the span of each prototype is
defined, SURF points within this span are matched with the prototype image and then
Homography transform is once again applied while considering the newly matched SURF points;
thus eliminating noisy detection/s of prototype. While the same process is repeated with each
cluster, a novel centroid based algorithm for merging repeated detections of same prototype
instance is used. Carrying the benefits of SURF and Homography transforms, the algorithm is
capable of detecting all prototype instances present in scene image, irrespective of their scale and
orientation. The complete algorithm has also been integrated into a desktop application, which
uses camera feed to report the real time count of the prototype in the scene image
131
Title : Rotor Angle Estimation and Coherency Detection of
Synchronous Generators
Author(s) : Gudla Ajay Kumar
Roll No : 13104049
Abstract:
Power system is the biggest, interconnected and complex system in the world. Since the demand
for power is increasing day by day, synchronous generators are forced to operate near the
stability limit. Hence, a fault or disturbance in system can cause the generators to step out of
synchronism, which may cause instability of the entire system. It is very important to access
stability of synchronous generators. In this thesis, we are concerned about angular stability only.
To access that, rotor angle should be monitored in real time. Since the power system is big and
interconnected, model reduction should be done for doing security analysis, transient stability
analysis etc. Coherency is a popular concept, that can be used for model reduction. To identify
coherent groups of generators, as a part of model reduction, we can use rotor angle of a generator
as a basis. But, rotor angle is not measurable in the system. Hence, the proposed work develops a
method to estimate the rotor angle in both transient and steady-state. With the advent of PMUs,
we are able to estimate the rotor angle in real time. A PMU is placed at the generator bus and it
measures bus voltage and branch current phasors. By using the PMU measurements and eld
current from machine, the proposed algorithm estimates the rotor angle. In the present work, 7th
order synchronous machine model is used. After disturbance occurs, in sub-transient state,
damper winding observer method estimates the rotor angle. In transient state, this method
switches to Kalman lter method. The proposed algorithm is tested on several standard systems
and implemented using hardware PMU interfaced with eMEGAsim real time digital simulator.
132
Title : Sliding Mode Control Strategies for Robotic Systems
Author(s) : Mitra Aritra
Roll No : 13104019
Supervisor(s) : Behera Laxmidhar
Abstract:
In this thesis, we investigate the feasibility of applying Sliding Mode Control (SMC), one of the
best known robust control techniques, to two applications - Targeted drug delivery inside the
human body using a fer- romagnetic micro-robot steered by a MRI, and Modeling and Control of
a 4 Degree of Freedom (DOF) Barrett Whole Arm Manipulator (WAM). It is extremely difficult
to obtain exact mathematical models of each of the systems under study, leading to unavoidable
modeling errors. The problem statement for both the applications can be formulated as trajectory
tracking problems of nonlinear dynamical systems with uncertainties. This motivates the use of
SMC. First, we develop an Adaptive Fuzzy Sliding Mode Control (AFSMC) scheme for the MRI
based drug delivery system, based on a highly nonlinear model available in literature. Simulation
results illustrate that the proposed technique is able to reject disturbances, and achieve perfect
tracking of the micro-robot along the centre line of a blood vessel. The ‘chattering’ phe-
nomenon observed in conventional SMC is completely eliminated. Another significant merit of
this framework, is its ability to estimate the dielectric density of blood on-line. Comparisons are
drawn with a state-of-the-art backstepping approach. Next, we develop a rigid body model of a 4
DOF Barrett WAM using the recursive Newton-Euler technique. This is a major contribution as
Bar- rett does not disclose its dynamic model. The precision achieved in rigorous trajectory
tracking experiments performed in the joint space, validates the accuracy of the developed
model. A variant of SMC, known as Nonsingular Fast Terminal SMC (NFTSMC), which
guarantees fast finite-time convergence of the error trajectories to zero, has also been proposed
for control of the WAM. We derive an analytical expression for the error settling time and
demonstrate that the NFTSMC indeed guarantees faster performance as compared to a standard
Nonsingular Terminal SMC (NTSMC). A detailed comparative study of simulation and
experimental results is presented. Finally, aiming to integrate the robustness of SMC with
optimal control theory, we integrate conventional SMC with Adaptive Dynamic Program- ming,
to obtain a novel optimal sliding mode control framework. We present the general design
procedure, stability analysis, and simulation results for both regulation and tracking problems,
the latter being
133
Title : Vision Based Navigation of a Quadcopter using a single
camera
Author(s) : Shree Atulya Shivam
Roll No : 10327172
Supervisor(s) : Behera Laxmidhar
Abstract:
Control of micro aerial vehicle is a major problem in indoor environments due to lack of reliable
position sensors. While outdoor applications can use GPS for reliable localization, working
indoors will require the use of Laser scanners or vision. If the indoor scene is not known to a
robot the task of mapping new areas also becomes a necessity. The two processes are combined
and run together in a framework of Simultaneous Localization and Mapping. Our work is
focused on using cameras for the task of SLAM in an indoor environment. Vision based
techniques though less accurate than Lasers have the potential of providing a low cost framework
for navigation. Two techniques for obtaining position feedback have been used in this work 1]
Localization using a known pattern in the scene and 2] Localization using a monocular SLAM
framework on an unknown and unstructured scene. For mapping the scene a probabilistic
framework has been implemented which can provide a semi-dense depth map of the
surroundings. Finally the localization algorithm has been implemented on a quadcopter and
experiments on hovering and waypoint-following demonstrated in an indoor lab based
environment.
134
Title : Voltage Stability monitoring using Lyapunov Exponent
Author(s) : Mukherjee Mriganka
Roll No : 13104082
Abstract:
In modern power systems, maintaining stability is a big challenge for the operators. Voltage
stability is a major stability problem encountered in modern power system. For a number of
recent power system blackouts, voltage instability has been found to be the main cause. Based on
the time frame of events, voltage stability problems can be classified into long-term and short-
term. For short-term stability problems, the control actions are mostly automatic, without much
scope for operator intervention. For long-term stability problems, usually, the operator gets
sufficient time to take preventive actions.The present work investigates the feasibility of using
Lyapunov exponents for early detection of any impending voltage stability problem. It is well
known that, near the point of voltage instability, the bus voltages may exhibit a chaotic behavior.
A number of indices proposed in the literature are based on the fact that, near the point of voltage
collapse, the system may go through a bifurcation. Lyuapunov exponent is a widely used index
to identify the chaotic behavior of a dynamic system. In this thesis, a thorough investigation is
done on the feasibility of using Lyapunov exponent for early detection of voltage stability
problems. Phasor measurement units (PMUs) are increasingly being deployed in modern power
systems. The main advantages of PMUs compared to conventional measurements are their
higher accuracy, higher reporting rates, and time-synchronization capability. The present work
explores the feasibility of using PMU measurements for monitoring short-term and long-term
voltage stability problems in power systems
135
Title : Power Quality Improvement and Power Capacity
Enhancement of Power Plant for Rural Telecom
Applications
Author(s) : Maji Abhishek
Roll No : 13104004
Supervisor(s) : Mishra Santanu Kumar&Joshi Avinash
Abstract:
This thesis describes a complete power plant architecture for rural telecom exchanges where grid
power is unreliable. The power plant is designed to work with one, two, or three phase input in a
seamless manner without any de-rating of the output power. The operational input of the
converter is between 90 V (rms) during single phase to 480 V (rms) during three phases. In order
to enforce unidirectional power flow and seamless phase selection operation, a four-leg diode
bridge is proposed for the input stage. The use of a four leg diode bridge to enforce automatic
phase selection in three phase power application is a challenge because the output of the diode
bridge has a wide voltage variation. In order to clip this voltage variation within an acceptable
limit, a clamp-enabled converter follows this bridge. The clamp-enabled converter can be either
step-up or step-down type depending on the input operational voltage range. The output of the
clamp-enabled converter is cascaded to a conventional isolated step-down converter. In this
paper, a full-bridge converter with current doubler is used. Passive power factor improvement
techniques are used to improve the power factor during three-phase, two-phase, and single-phase
operation. The proposed system is verified using a 1 kW prototype. Average current mode
control technique is adopted for the proposed power plant to obtain regulated voltage at the
output. To increase the current sourcing capability of the entire system and simultaneously
reduce the switch stress, multi module paralleling is performed. Single wire autonomous current-
sharing paralleling technique is employed for current-mode-controlled dc power modules. The
proposed control circuit is designed by the three-loop control method. A design example of two
400 V/ 48 V, 25 A parallel modules is set up and experimental recordings verify the performance
of current sharing.
136
Title : Modeling, Analysis, and Implementation of a Hybrid MOC
for Solar PV Applications
Author(s) : Jamal Ishrat
Roll No : 13104055
Abstract:
Converters having high conversion ratio are preferred in renewable applications because they
allow use of smaller solar panels to achieve higher output voltage. This leads to better
performance of a solar-PV system as the power extraction is less susceptible to partial shading. A
Current-Fed Switched Inverter (CFSI) has a very high conversion ratio of the order of 8-10,
under practical scenario. The CFSI is hybrid Multi Output Converter (MOC), i.e., it is capable of
producing AC and DC outputs, simultaneously, from a single input DC source. All these
properties make it an excellent candidate for solar-PV applications in residential renewable
systems. In this thesis, modeling, control, and usage of CFSI in a renewable nanogrid is
presented. The steady-state and dynamic model of the converter are studied in detail.
Conventional circuit averaging technique and energy conservation technique are used to derive
these models. The AC/DC control-to-output response, output impedance characteristics, and
audio susceptibility are derived. A lab prototype is designed to verify the models. The results are
verified using PSPICE simulations and experimentally using a Frequency Response Analyzer
(FRA). The output DC link of a CFSI is regulated using analog as well as digital controllers. The
theory of controller implementation is derived and verified using conventional Bode plot
method. The robustness of the feedback design is tested by testing the converter response to a
step change in load and reference voltage. A soft-start technique for CFSI is discussed and
implemented in digital domain. The application of a CFSI to a residential renewable nanogrid is
verified using a lab scale prototype. The CFSI is designed to track Maximum power point (MPP)
with a battery or a resistive load at the DC link. MPP is verified using three methods, (a) P & O
method with a fixed DC link voltage, (b) P & O method with resistive load at the DC link, and
(c) Input voltage control. In all the three methods, the MPP is tracked properly and maximum
power from solar-PV is transferred to the load. The merits and de-merits of all the three methods
are discussed.
137
Title : Design and analysis of ultra-lift LUO converter based PFC
circuits
Author(s) : Srinivas Vanaparthi Venkata
Roll No : 13104153
Abstract:
This thesis studies the feasibility of using a higher order converter with high gain to implement a
power factor correction rectifier. Ultra-lift Luo converter is a good choice for this application as
it offers very high gain at lower duty cycles compared to conventional converters, e.g., boost
converter. Based on Ultra-lift Luo converter, two new PFC (power factor corrected) circuits are
proposed and analyzed. The first PFC circuit using this higher order converter is a bridge-less
topology. It has unidirectional power flow capability. The proposed circuit uses two less passive
components (one inductor and one capacitor) compared to the state of the art found in literature.
The circuit is operated in DCM. The proposed circuit is validated using PSPICE simulation. The
second PFC circuit is obtained by replacing the converter at the output of a diode bridge rectifier
(in conventional PFC network) with an Ultra-lift Luo converter. Current sensing is avoided by
operating the converter under DCM. Steady state and small signal models of the network are
derived and voltage feedback regulation is implemented. The open loop operation of this
proposed PFC network is validated by using an experimental prototype.
138
Title : Series Active Filter for Rectifier DC Link Current
Author(s) : Pakhira Ambar
Roll No : 13104012
Supervisor(s) : Sensarma Partha Sarathi
Abstract:
High power rectification (AC to DC) applications are invariably beset with problems of low-
order even harmonic components in the rectified output variable. For applications which require
a current input, e.g. HVDC converters, current source inverters, traction front-ends etc.,
attenuation of these harmonics are of utmost importance. Thus, in order to realize a current stiff
DC link, the filtering requirement is stringent. Usually, to avoid increase in system complexity,
an inductive filter is used, but there is a consequent reduction in efficiency and power density. To
improve both these critical operating metrics (efficiency, power density) this thesis focuses on a
series active filter scheme to provide additional attenuation, over and above the passive inductive
filtering. This brings about a large reduction in the overall filter size, with a possibility of
increase in efficiency. As per analytical investigations and numerical simulation, the inductance
of a standard inductive filter could be reduced by a factor of 10, without even slightly affecting
the low-order harmonics in the rectifier output current. The proposed active filter senses the
rectifier output voltage and extracts the low order (even) harmonic components from the sensed
voltage using a digital filter. It then injects a voltage in series with the inductor in order to ensure
negligible low order harmonics in the voltage appearing across the inductor. Thus a stiff D.C.
current link can be achieved with a smaller inductor. The control mechanism consisting of a
single loop was implemented using FPGA. Simulation results show negligible lower order ripple
in the D.C. link current. A lab prototype series active filter for a 700 VA rectifier was built using
power MOSFETs as switches. The switching frequency was selected to be 10 kHz for low
switching loss while not sacrificing control speed to a greater extent. Input to the rectifier was
single phase 50 Hz 110V AC. This lab prototype was used to validate the analytical and
simulation results.
139
Title : Design and Development of Single Phase Dynamic Voltage
Restorer
Author(s) : Meena Amit Kumar
Roll No : 13104013
Abstract:
Dynamic Voltage Restorer (DVR) is a custom power device used in electrical distribution
system for power quality improvement. The main application of DVR is for voltage
compensation of sensitive loads against voltage disturbances like voltage sag and voltage swell
in distribution lines. It is a series connected device and is able to compensate voltage sag and
voltage swell by injecting a voltage with help of series transformer. The injection of an
appropriate voltage component in the event of a voltage disturbance requires a certain amount of
real and reactive power. Conventionally DVR consists of an energy storage device which
supplies the required power over the limited duration of the sags. Large and long duration sags
lead to heavy financial investment in energy storage unit. To overcome this limitation, a single
phase back-to-back converter based DVR is implement in this thesis,which eliminates energy
storage requirement. Two converters are connected through common dc link. In the event of
voltage sag, front-end converter acts as active rectifier and other converter acts as inverter,
injecting the voltage in series with grid voltage. Real power flows from rectifier to inverter. In
case of voltage swell condition, the operation of two converters interchanged and energy is fed
back to the source side converter from the load side converter. Modelling of both converters is
done to determine the behavior of the system. For unity power factor operation of front-end
converter, Phase Locked Loop is used which track the grid frequency and phase. Both converters
are operated in closed loop. Front-end converter maintains the dc-link voltage and current drawn
by DVR to unity power factor. Other converter maintains the injected ac voltage to the desired
value. The DVR is designed for 5kVA load and source voltage in the range of 185V to 265V.
DVR along with designed controllers are tested by simulation on MATLAB/Simulink. To
validate the performance of the DVR, a laboratory prototype of 5KVA is developed. Control
algorithms are implemented in TI make DSP TMS320F2808.The prototype is tested on both
linear and non-linear loads.Performance of the prototype is also validated for step change in
load.The results obtained from simulation approach and hardware are found to be in agreement.
Various protection schemes such as, protection against under-voltage, over-voltage and
overloading are implemented in the developed prototype.
140
Title : Study of Silicone Rubber based Composites with Nano- sized
Fillers
Author(s) : Kumar Rakesh
Roll No : 10327580
Supervisor(s) : Gupta Nandini
Abstract:
Silicone rubber (SR) has emerged as a material of choice for housings in outdoor insulation in
high voltage transmission. The surface hydrophobicity of SR based insulators makes them ideal
for use as outdoor insulation especially in polluted environments. Research in nanodielectrics in
the last decade has shown that addition of nanosized fillers into bulk polymers often improve
their electrical and non-electrical properties significantly. Such properties include permittivity,
breakdown strength mechanical strength, etc. In this work, we study the effect of incorporating
barium titanate nanofillers into room temperature vulcanized silicone rubber (RTVSR). Nano-
sized barium titanate (BaTiO3) particles are used as fillers in the RTVSR matrix with volume
fractions of 0.5%, 1%, and 2%. While preparing samples, mechanical mixing and ultrasonic
mixing is used to reduce the agglomeration of nano fillers. Scanning Electron Microscope (SEM)
micrographs are used to thoroughly examine the dispersion of the nanofillers in the base matrix.
The findings are also evaluated against silicone rubber composites filled with alumina instead of
barium titanate. One of the most important issues in outdoor insulation is tracking specially in a
polluted environment. Surface contamination of the insulator, and the consequent leakage
currents, can cause operational problems in electric power utilities. Tracking on the surface of
the insulator leads to degradation and even complete failure of the insulator over time. In this
work, we study the effect of incorporation of nano-fillers on the propensity of the RTVSR to
tracking and associated degradation. Inclined Plane Test (IPT) for erosion as per the IEC60587 is
performed. Additionally, surface degradation tests are performed to estimate how the addition of
nano fillers increases the resistance to degradation due to surface discharges.
141
Title : Design, Simulation and Development of a Multifunctional
Active Power Filter for Power Distribution System
Author(s) : Kaka Bharadwaj
Roll No : 13104028
Abstract:
With increased penetration level of digital technology in power system, along with the growth of
non-linear and power electronic loads, the current drawn from the grid is becoming far from
sinusoidal. The non-sinusoidal currents drawn electrically pollute the power system resulting in
poor power quality in addition to increased losses and stress in the transformers, lines and
equipment. Furthermore, there are problems of voltage sag/swell and unbalancing in
voltage/current. Thus there is a need to have a power quality conditioner/active power filter at
appropriate voltage level to compensate for load current harmonics as well as voltage sag/swell
and voltage unbalance. If necessary, the active power filter can also be additionally designed to
compensate for load reactive power requirement. The present project aims at an integrated
design, simulation and development of an active power filter for 415V systems. The topology for
load current compensation & voltage sag/swell mitigation is simulated with non-linear,
unbalanced and induction motor loads. Subsequently, a three phase 415V active power filter
(prototype) is developed in the laboratory. The proposed active filter enables • Compensation of
load current harmonics • Compensation of load reactive power • Compensation of unbalance in
PCC voltages and load currents • Compensation of balanced/unbalanced voltage sag/swell and
voltage flicker
142
Title : Design and Modelling of VSC-HVDC Transmission System
Having Wind Farm
Author(s) : Roy Nirjhar
Roll No : 13104093
Supervisor(s) : Singh Sri Niwas
Abstract:
Over the last few years, installation of Voltage Source Converter based High Voltage Direct
Current (VSC-HVDC) transmission systems have been increased considerably; due to lesser
space requirements, better power quality, reactive power controllability, etc. Several control
methods are available, though only a few are practically viable for grid integration. The work
presented in this thesis investigates such a design method for the controllers of VSC-HVDC. The
simulation results show that the designed controllers work well whether isolated from grid or
integrated to grid. When connected with a wind farm and the grid, the outputs follow the grid
codes for renewable energy generation. However, inclusion of fuzzy logic controllers (FLCs)
improves the performances of the integrated system. A change in control method is proposed,
where the FLCs are used in such a way that the complexity of the system reduces significantly
and the performances improve. The advantages of the conventional controllers are still present,
and merits of FLCs add on top of that as both controllers are used together in the control system.
Further, the exploration with FLCs may be done to achieve better results for isolated or
integrated power systems.
143
Title : DSP based Implementation of Induction Motor Control with
Estimation of Rotor Time Constant
Author(s) : Kandlagunta Rahul Chowdary
Roll No : 13104064
Abstract:
Induction Motor is made to operate as variable frequency drive with Scalar Control, Vector
Control techniques etc. DC Motor characteristics are replicated in Induction Motor using Vector
Control technique. Vector Control gives good dynamic performance characteristics as flux and
torque component currents are decoupled. In this thesis, Indirect Vector Control with rotor flux
oriented synchronous reference frame is chosen. The slip speed required for synchronous speed
calculation depends on the rotor time constant. Rotor Time Constant varies with temperature and
this variation is estimated for proper rotor flux orientation. Rotor Time constant is estimated
using Reactive Power Model Reference Adaptive Scheme. Simulation analysis is carried out in
MATLAB SIMULINK to evaluate the performance of control schemes. A Lab prototype has
been developed for the experimental verification. Controllers are implemented in
TMS320F28335 DSP experimental kit. Hardware results are presented and they are compared
with the simulation results with good agreement.
144

Mtech Thesis Abstract 2015

Uploaded by

Copyright:

Available Formats

Mtech Thesis Abstract 2015

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Mtech Thesis Abstract 2015

Uploaded by

Copyright:

Available Formats

M.TECH.

THESIS ABSTRACT 2015

2 Microelectronics,VlSI &Technology Display 7-8

3 RF Microwaves & Photonics 9-10

4 Power &Control 11-12

Sr Title Supervisor Page

1 Video Stabilization And Object Tracking Using Affine- Gupta Sumana 13

6 People counting in high density crowds from still images Venkatesh KS 18

8 SBL based joint sparse channel estimation and maximum Jagannatham 20

18 GLRT based target detection in monostatic MIMO RADAR Jagannatham 30

28 Online Tracking Of Low Rank Matrices From Partial

30 On Optimal User Pairing and Opportunistic Interference Chaturvedi A K 42

35 On Redundancy Rate of SWLZ and FDLZ Algorithms and Bansal Rakesh K 47

37 Subspace Based Domain Adaptation for Visual Recognition Hegde Rajesh 49

40 Regularity Flow Inspired Target Tracking in FLIR Imagery Venkatesh K S 52

41 Hybrid Structured Peer to Peer Live Video Streaming Singh YN 53

47 Localization enhancement of wireless sensor networks Rajawat Ketan 59

48 Fast CFO estimation in OFDM systems using golden section Vasudevan 60

55 Fault Tolerant Node Disjoint Multipath Routing In Manets SinghYN 67

56 PAPR Reduction In OFDM System By ZADOFF-Chu Vasudevan 68

65 Shape reconstruction for ISAR imaging Naik Naren 77

66 Marker Less Hand Detection And Gesture Recognition From K S Venkatesh 78

68 Precoding with time-domain feedback in MIMO-OFDM Vasudevan 80

70 On Techniques to Detect Malicious Users in Cooperative Banerjee Adrish 82

Microelectronics,VlSI &Technology Display

Sr Title Supervisor Page

7 Study of suitability of four imidazolin-5-one derivatives for Iyer S Sundar 90

2 Studies on Multilayer Substrate Integrated Waveguide Slot Biswas 106

4 Design of X-Band Power Amplifier based on GAAS-PHEMT Biswas 108

5 Design of X-Band low phase-noise oscillator and comparative Biswas 109

6 Design of power amplifier Biswas A 110

7 Design of Broad Band Bias-T on Alumina Substrate Das Utpal 111

8 Design of broadband Bias-T on GAAS substrate Das Utpal 112

9 Design of Planar Resonator Sensor for RF Testing of Liquids Akhtar M J 113

10 Design of Multi Slot Vivaldi Antenna with Improved Akhtar M J 114

11 Continuously Tunable Multi-wavelength Ultra-Fast Fiber Ring Kumar Pradeep 115

13 Fractal Geometry Inspired Polarization Insensitive Microwave Srivastava 117

15 Compact And Enhanced Bandwidth Polarization Harish A R 119

16 Design, Fabrication and Characterization of a Vacuum Gated Das Utpal 120

Sr Title Supervisor Page

12 Rotor Angle Estimation and Coherency Detection of Chakrabarti 132

21 Study of Silicone Rubber based Composites with Nano- Gupta 141

For more details click here back

For more details click here back

In this thesis, we propose a framework for distributed dissemination of messages by leveraging

For more details click here back

For more details click here back

For more details click here back

For more details click here back

For more details click here back

For more details click here back

For more details click here back

In this thesis work, we consider the decentralized detection in cooperative multiple-

For more details click here back

For more details click here back

For more details click here back

Colorization refers to the process of producing a coloured image/video from a given

For more details click here back