|
|
Opportunistic
Multiuser Communication - Beamforming with Predictive Scheduling
Wireless networks with multiple users provide multiuser-diversity arising
out of multipath fading. To exploit this diversity, channel for each user
is tracked and the user experiencing the best SNR is scheduled for transmission
(considering downlink). However these gains are limited due to the slow
fading or little scattering environments and Viswanath et. al. proposed
artificially-induced channel-variations using multiple antennas in such
environments. This project discusses the multiuser diversity gains in
fading environment and suggests a predictive scheduling approach to enhance
the capacity gains. The Gauss-Markov model for the time-varying parameters,
as suggested by Sanghavi et. al. [4] is used to analyse the dynamics and
gains. ( Presentation)
|
|
|
Detecting worms through
de-centralized monitoring Worm attack is a large-scale denial-of-service
attack on the internet. The attack, because of its global propagation follows
some dynamic model. Its behaviour is based on a simple epidemic model with
a slow start phase during which infected hosts increase exponentially with
a positive infection rate. Based on this model Towsley et. al. put forth
the idea of “detecting the trend, not the rate” of monitored worm
scan traffic and employed a Kalman filter to detect worm propagation. They
also proposed a centralized worm monitoring architecture where monitors
distributed over the internet gathered data on worm activities and communicated
it to a Central Malware Warning Centre. This report suggests a two-level
hierarchy for the monitoring system. The first-tier monitors collect the
scan data at the ISP level. And second-tier monitors share the estimates
to detect the trend across ISPs i.e. globally. The first-tier monitors estimate
the trend by employing LMS algorithm in a distributed manner. And fast-distributed-linear-averaging
techniques are employed to make ISPs come to consensus about the global
trend. |
|
|
Methods of Quadratic Phase Coupling
(QPC) Estimation This project was a part of my undergraduate
thesis at Netaji Subhas Institute of Technology (NSIT), Delhi under the
supervision Dr. Harish Parthasarathy, Department of Electronics and Communication
Engineering, NSIT. The project involved understanding Higher Order Statistics,
implementing various techniques of estimating 3rd order cumulants and
bispectrum; understanding the Quadratic Phase Coupling problem and its
application to non-linear system identification, array processing and
image processin; implementing parametric (AR) and eigen-subspace based
algorithms (MUSIC and ESPRIT) for Quadratic Phase Coupling (QPC) Estimation.
Extending Dr. Parthasarathy's PhD Thesis work, the ESPRIT algorithm was
improved and made computationally efficient, thereby reducing number of
MIPS, and rendering it for application to practical scenarios. (Bachelor
Thesis) (Presentation)
|
|
|
|
|
|
Estimating
Power Spectra of Natural Images Statistics of natural images
have been found to follow particular regularities. Seminal studies (Burton
and Moorhead 1987, Field 1987, 1994, Tolhurst et al 1992) have observed
that the average power spectrum of natural images falls with a form 1/f^a
with a ~ 2. Related studies found a bias in the distribution of orientations,
illustrated in the power spectra of natural images. In real-world images,
including both natural landscapes and man-made environments, vertical and
horizontal orientations are more frequent than obliques (van der Schaaf
and van Hateren 1996) i.e. the power spectra is anisotropic. Thus, a more
complete model of the mean power spectra (using polar coordinates) can be
written as A_s(\theta)/f^{a_s(\theta)}in which the shape of the spectra
is a function of orientation. The function A_s(\theta) is an amplitude scaling
factor for each orientation and a_s(\theta) is the frequency exponent as
a function of orientation. Both factors contribute to the shape of the power
spectra. This model is needed when considering the power spectra of natural
images (Baddeley 1997). The goal of this project is to investigate and
study the power spectral density of an ensemble of natural images and estimate
a model for the same |
|
|
Wavelets,
Compression and Denoising Wavelet expansions and wavelet
transforms have proven to be very efficient and effective in analyzing
a very wide class of signals and phenomena. Wavelet expansion allows a
more accurate local description and separation of signal characteristics.
While Fourier coefficient represents a component that lasts for all time,
a wavelet expansion coefficient represents a component that is itself
local and is easier to interpret. The goal of this project is to investigate
the discrete wavelet transform (DWT) and its application to signal and
image compression and de-noising, both in 1-D as well as 2-D.
|
|
|
Content Based Coding
Of Face Images The wavelet transform has recently emerged
as a promising technique for signal pro- cessing applications due to its
flexibility in representing the non-stationary signals. The wavelet representation
provides a multiresolution/multifrequency expression of a signal with
localization in both time and frequency. It decomposes a given signal
into a set of coefficients associated with multiscaled wavelets. This
property is very much desirable in image and video coding applications,
as coding schemes and parameters can be adapted to the statistical properties
for each of wavelet coefficients. Wavelet transform based image and video
coding techniques have the advantage that they are free from the blocking
artifacts due to the nature of its global decomposition. Among all the
schemes, content-based image and video coding techniques provide the best
image quality at low bit rates. In these techniques, knowledge of the
underlying image and video is exploited to achieve the best results. One
example of this could be the content-based coding of face images. Face
images form the important database in the police departments, banks, security
kiosks, and they are found in abundance in day-to-day life. In these databases,
the important content of course is the face region. The image coding techniques
which masks the face regions for discriminative quantization will be of
most importance. These approaches can incorporate human visual system
(HVS) aspects into the coding schemes, as the final viewer is going to
be human. (PPT
and results)
|
|
|
|
|
|
Speech
Modelling This project was a part of winter training done
at NSIT under the supervision Dr. Harish Parthasarathy. The report discusses
the algorithms for Optimum Linear Prediction and Filtering. Levinson-Durbin
algorithm is implemented and its performance as a tool for System Identification
is analyzed. The algorithms for Adaptive Filtering, in particular, Recursive
Least Squares is implemented and its performance as tool for System Identification
is analyzed. Both the approaches are then employed for Linear Predictive
Coding and Speech Modelling. The comparisons and inferences therein are
captured in the report.
|
|
|
FSK Caller
ID This report describes the algorithmic as well as implementation
details of FSK based Caller Line Identification Presentation (CLIP) service.
The scope of the document is limited to CLIP simulator which includes
Transmitter only. This document covers following areas: FSK modulation
as per ITU-T V.23, CLIP protocol as per ETSI Recommendation ETS 300 659
and ETS 300 778 and Signal processing methods used in the design of CLIP
simulator.
|
