Title: Radios for the Internet of Things

Speaker: Prof. Gaurab Banerjee, ECE Dept., IISc

Date: 22 August 2014 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: The Internet of Things (IoT) is expected to include 50 Billion devices by the year 2020. As of recent reports,
there may be 5000+ types of devices in such a massively distributed system, with projected revenues in the range
of $500B to $1T for such devices. Since most of the embedded devices will be of small form-factors, consume low
power, and transmit and receive data using wireless links, this presents a tremendous opportunity for chip designers
and manufacturers.

In this talk, I will provide a brief overview of how the last few generations of radios have evolved to the point
that today, the term broadly includes a transceiver, that is supplemented by several microprocessors, memory
and signal processing engines. The radios of tomorrow will have elements of cognition built in such a way that
in addition to communication, they will provide intelligent power and spectrum management. I will identify three
key technologies of today (low power radios, cognitive radios and millimeter-wave radios), which will be the key
ingredients necessary to construct radios for the internet of things. Next, I will discuss a few architectures
and ingredients that are being studied in my research group, which can be the key ingredients in radios, built
for the internet of things.

Title: Harnessing bursty sources and resources for interference management

Speaker: Prof. Suhas N. Diggavi, Dept. Electrical Engineering, Univ. of California Los Angeles (UCLA)

Date: 01 August 2014 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: Managing interference is one the fundamental challenges for current and future wireless networks.
One of the tools extensively used in deployed networking systems is the use of feedback to learn and adapt
to the unknowns, including demands, network conditions etc. In this talk we explore the use of feedback
for wireless interference management when there is burstiness in the wireless traffic and when feedback
resources are intermittently available. Burstiness occurs in many realistic scenarios. For example, wireless
traffic is bursty owing to mechanisms in the medium-access control layer, the networking protocol in the
network layer etc. Resources for feedback could also be bursty (or intermittent) for example due to
opportunistic use of side-channels. In this talk we develop coding schemes that can harness bursty sources
and resources which we can show are approximately optimal.

Parts of this talk are joint work with Can Karakus, I-Hsiang Wang, Shaunak Mishra, Changho Suh
and Pramod Viswanath.

Title: Wireless Power Transfer in the Age of Internet of Things

Speaker: Prof. K.J. Vinoy, ECE Dept., IISc

Date: 18 July 2014 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: Internet of things aims at providing universal connectivity. While short distance communication
is indeed possible with little or no energy at the terminal, most sensing devices and/or their management
require a power source. Practical sensors require anywhere between 50uW to about 10mW for their
operation. Present day batteries have limitations in terms of their cost, size, lifetime and energy capacity.
Therefore, to extend the operational life of such sensors, various energy harvesting schemes have been
attempted. RF energy harvesting schemes have the unique advantage that these circuits may work as
receptors even for wireless transfer of power. Challenges and achievements in practically exploiting
RF energy harvesting to operate wireless terminals in various scenarios will be discussed during this talk.
Having identified low energy density available in the ambient as the main challenge, we are presently
investigating a possibility for high Q components and circuits to overcome this. Most harvesting circuits
presented are developed in collaboration with Dr. TV Prabhakar, DESE.

Title: Circuits, brains, and crowds: Reliable inference in systems with unreliable components

Speaker: Prof. Lav Varshney, Univ. of Illinois, Urbana-Champaign

Date: 16 May 2014 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: When considering information processing systems like circuits, brains, or crowdsourcing systems,
components such as logic gates, synapses, or people may be noisy. In this talk, I present three vignettes
of work on using coding-theoretic ideas to achieve reliable performance of such systems in the face of
unreliable parts. In particular, I discuss how to build memories that can store information reliably forever
despite being constructed from noisy registers and gates; associative memory models of biological brains
that can reliably retrieve information despite having noisy synapses and facing noisy queries;
and crowdsourcing microtask platforms that can reliably perform fine-grained classification despite
unreliable and unskilled workers.

Title: Recent Advances in Photonic Integrated Circuits for Communication and Sensor Applications

Speaker: Prof. T. Srinivas, ECE Dept., IISc

Date: 25 April 2014 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: Micro ring resonators and photonic crystals are emerging as the structures of choice to implement optical functions required for
communication and sensor applications, replacing conventional integrated optic devices like directional couplers, Mach Zhender
interferometers etc. Many functions like power dividers, filters, mux/demux could be implemented more compactly with these structures.
The talk reviews the existing technology and takes up on-going work in our laboratory, including fabrication studies. The devices of
interest include: ordinary ring resonators and arrays, photonic crystal ring resonators, MEMS coupled ring resonators and MEMS
coupled photonic crystal ring resonators. Applications include power dividers, filters, mux/demux, pressure, acceleration and force sensing.

Title: Design and Deployment of Multihop Wireless Networks for Sensor Interconnection

Speaker: Prof. Anurag Kumar, ECE Dept., IISc

Date: 11 April 2014 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: In the context of Cyber Physical Systems and the Internet of Things, there is a growing need for the
design and deployment of multihop wireless relay networks for connecting sensors to a control center. We
have considered the following problem. Given the locations of the sensors, the traffic that each sensor
generates, the quality of service (QoS) requirements, and potential relay locations, place a small number
of relays at some of the potential locations so as to realise a multihop wireless relay network that achieves
the desired QoS for sensor generated traffic. In this talk, we will provide an overview of our research
on algorithms for addressing such a problem. We have also developed SmartConnect, a design and deployment
tool that employs our algorithms, in a field interactive, iterative approach, with model based network design,
field evaluation, and relay augmentation performed iteratively until the desired QoS is met. We will provide
an overview of the design choices made in SmartConnect and describe the experimental work that led to these
choices. We provide results from some experimental deployments. Finally, we conduct an experimental study of
the robustness of the network design over long time periods (as channel conditions slowly change), in terms of
the relay augmentation and route adaptation required.

Title: Active Shapes

Speaker: Prof. Chandra Sekhar Seelemantula, EE Dept., IISc

Date: 28 March 2014 (Friday)

Time: 4:00pm

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: We address the problem of biomedical image segmentation using shape constraints.
The shape model is user-specified and the segmentation problem is solved using an
affine registration-type approach. The mathematical formulation is generic and capable
of handling a variety of convex or concave shapes. The optimization framework turns out
to be independent of the shape complexity. We present examples on some frequently
encountered classes of biomedical images (fundus images, magnetic resonance images,
mammograms, western blot/gel electrophoresis images, microscopy images, etc.). We show
that the proposed method is robust to occlusion, partial loss of structure, and Gaussian
as well as Poisson noise distortions. We also extend the formulation to perform segmentation
by shape learning, starting from a set of images manually segmented by an expert.

Title: Distributed Sequential Detection Algorithms for Spectrum Sensing

Speaker: Prof. Vinod Sharma, ECE Dept., IISc

Date: 14 March 2014 (Friday)

Time: 4:00pm

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: In a Cognitive Radio (CR) system the CRs need to sense the channel to check if it is being used by a primary user or not.
The CRs use the channel only when it is not being used by the primaries. The sensing needs to be done at a very low SNR (-20dB).
Sensing reliably at such low SNRs in a wireless channel is particularly challenging because of time varing fading and shadowing.
Thus sensing simultaneously by multiple CRs located at geographically distributed locations and detecting based on these
observations is a feasible solution. In this talk I will present some of the distributed sequential detection algorithms
developed in our group and present their analysis.

Title: Secure Compute-and-Forward Using Nested Lattice Codes

Speaker: Prof. Navin Kashyap, ECE Dept., IISc

Date: 31 Jan. 2014 (Friday)

Time: 4:00pm

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: The wireless two-way relay model is a key primitive or building block in
physical-layer network coding schemes. In this model, a relay helps two users
to exchange their messages, which are assumed to lie in some finite Abelian group G.
The compute-and-forward strategy has two phases - a multiple-access (MAC) phase
and a broadcast phase. In the MAC phase, the two users transmit real-valued signals
(satisfying some power constraint) that encode their respective messages, and the relay
receives the superposition (real-valued sum) of the two signals with some Gaussian noise
added. From the received signal, the relay computes the sum, in the group G, of the users'
messages, and in the ensuing broadcast phase, it forwards the computed sum to the
two users. Nazer and Gastpar (2011) showed that the two users can reliably exchange
their messages at any rate up to (1/2)*log(1/2 + SNR) using nested lattice codes.

In this talk, we impose an additional security requirement: the relay is untrusted, so
it should get little or no information about each individual user's message beyond what is
obtainable from the sum it can compute. We devise novel coding schemes involving nested
lattice codes and randomization by well-chosen probability distributions on the lattice points.
The choice of probability distribution determines the strength of the security guarantee.
We show that we can have reliable and perfectly secure exchange of messages at any rate
up to rate (1/2)*log(SNR) - log(2e) using our coding scheme. If we relax perfect security
to strong information-theoretic security, then rates up to (1/2)*log(1/2 + SNR) - (1/2)*log(2e)
are achievable.

This is joint work with Shashank Vatedka and Andrew Thangaraj.

Title: Information-Theoretic Tools From/For Cryptography

Speaker: Prof. Manoj Prabhakaran, University of Illinois, Urbana-Champaign, USA

Date: 17 Jan. 2014 (Friday)

Time: 4:00pm

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: In this talk I will describe certain information-theoretic tools developed for analyzing
cryptographic questions, which turned out to have applications in other (non-cryptographic)
information-theoretic problems as well. No background in cryptography will be assumed.
(Based on joint work with Vinod Prabhakaran and others.)

Title: Hybrid, Energy Harvesting, Opportunistic Wireless Systems

Speaker: Prof. Neelesh B. Mehta, ECE Dept., IISc

Date: 29 Nov. 2013 (Friday)

Time: 4:30pm

Venue: Golden Jubilee Hall, ECE Dept.

Abstract: Energy harvesting (EH) is an attractive and green solution to the
problem of limited lifetime of wireless sensor networks (WSNs). Unlike a
conventional node that dies once it runs out of energy, an EH node
harvests energy from the environment and replenishes its rechargeable
battery. We investigate opportunistic hybrid networks that comprise of
both EH and conventional nodes, and differ from both conventional and
all-EH WSNs. We propose two new and insightful performance criteria
called k-outage duration and n-transmission duration to evaluate these
networks. They overcome the pitfalls associated with defining lifetime,
which arise because the EH nodes never die but can occasionally run out
of energy, and capture the dynamic time evolution of hybrid networks,
which occurs because the conventional nodes irreversibly drain their
batteries. They account for the effects of insufficient battery energy
and channel fading. We develop two novel bounds for evaluating these
criteria. We also discuss distributed selection algorithms for implementing
opportunism in such networks.

Title: Optical and hydrodynamical effects in chiral nanostructures and their interplay

Speaker: Prof. Ambarish Ghosh, ECE Dept. and CeNSE, IISc

Date: 08 Nov. 2013 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Room ECE 1.08, ECE Dept.

Abstract: The ubiquitous influence of chirality in the biological world include the differential pharmaceutical activity of right
and left handed drug molecules, locomotion of micro-organisms through the rotation of helical (and therefore chiral)
flagella, and many more. My talk will focus on a system of artificial chiral nanostructures that can be actuated with
small magnetic fields to move through fluidic media similar to various bacterial species. It has been possible to achieve
extremely well controlled motion with these nanostructures, which can be used in various biomedical applications, and can
also serve as a model system to understand certain complex biological processes. Interestingly, these structures can be
engineered to have strong chiro-optical response, which can affect their dynamics in interesting ways.

Title: Local algorithms for large-scale network problems

Speaker: Prof. Rajesh Sundaresan, Dept. of Electrical Communication Engineering, IISc

Date: 25 Oct. 2013 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Room ECE 1.08, ECE Dept.

Abstract: Local algorithms are distributed iterative algorithms designed to work
on a network with a neighbourhood structure. Each node makes repeated
updates of its state based only on information gathered from nodes in
its neighbourhood. The algorithm designer must design update rules to
make the node states settle down, eventually, at certain desired values.
Some examples are price discovery in economics, spread of information or
influence in sociological networks, etc., and in electrical engineering
and computer science (EECS), access protocols in WiFi networks, consensus
reaching in autonomous systems, synchronisation of clocks of
interconnected devices, decoding of data bits in 3G mobile phones, network
optimisation problems, and several instances of some hard combinatorial
optimisation problems but under some parameter regimes.
We will discuss some of the aforementioned EECS examples and will attempt
to bring out the issues involved in designing good local algorithms.

Title: On finding a set of healthy individuals from a large population

Speaker: Prof. Chandra R. Murthy, Dept. of Electrical Communication Engineering, IISc

Date: 04 Oct. 2013 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Room ECE 1.08, ECE Dept.

Abstract: In this talk, we consider the problem of finding a given number of "healthy" items from
a large population containing a small number of "defective" items using nonadaptive group testing.
Such problems arise, for example, in the spectrum hole search problem for the cognitive radio (CR)
networks. It is well-known that the primary user occupancy (active set) is sparse in the frequency
domain over a wideband of interest. To setup a CR network, the secondary users need to find an
appropriately wide unoccupied (inactive) frequency band. Thus, the main interest here is the
identification of only a sub-band out of the total available unoccupied band, i.e., it is an inactive
subset recovery problem. Another example is a product manufacturing plant, where a small shipment
of non-defective (inactive) items has to be delivered at high priority. Once again, the interest here
is in the identification of a subset of the non-defective items using as few tests as possible

For this problem, we present mutual information based upper and lower bounds on the number of non-
adaptive group tests required to identify a given number of healthy items. We show that an impressive
reduction in the number of tests is achievable compared to the approach of first identifying all the
defective items and then picking the required number of healthy items from the complement set.
Further, in the nonadaptive group testing setup, we obtain upper and lower bounds on the number of
tests under both dilution and additive noise models, and show that the bounds are order-wise tight.
Our results are derived using a general sparse signal model, by virtue of which, they are also
applicable to other important sparse signal based applications such as compressive sensing.

Title: Spatial Modulation: Icing on the MIMO Cake

Speaker: Prof. A. Chockalingam, ECE Dept., IISc

Date: 27 Sept. 2013 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Room ECE 1.08, ECE Dept.

Abstract: Spatial modulation is a relatively new modulation scheme suited for
multiantenna (MIMO) communications. It allows the use of less number of transmit
RF chains than the number of transmit antennas without compromising on the spectral
efficiency. This reduces RF hardware complexity, size and cost in MIMO systems.
A novel aspect of this modulation scheme is that it conveys information in the
indices of the chosen antennas for transmission, in addition to information
conveyed through conventional modulation alphabets like QAM. The talk will introduce
spatial modulation, its relevance in MIMO systems with large number of antennas,
some of our recent results, and potential topics for future research. The talk
will also briefly highlight 'NAVA-Plus', a spatial modulation extension to 'NAVA',
a large MIMO system development project by DRDO and IISc.

Title: Music and symbolic dynamics: The science behind an art

Speaker: Prof. Shayan G. Srinivasa, Dept. ESE, IISc

Date: 30 August 2013 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Seminar Hall, ECE

Abstract: Music signals comprise of atomic notes drawn from a musical scale. The creation
of musical sequences often involves splicing the notes in a constrained way resulting in
aesthetically appealing patterns. We develop an approach for music signal representation
based on symbolic dynamics by translating the lexicographic rules over a musical scale to
constraints on a Markov chain. This source representation is useful for machine based music
synthesis, in a way, similar to a musician producing original music. We will present several
examples of artificial music synthesis inspired by the south Indian classical music system.

Title: Bio-Molecular Sensing: Present and Future

Speaker: Prof. Manoj Varma, ECE Dept. and CeNSE, IISc

Date: 26 July 2013 (Friday)

Time: 4:00pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Seminar Hall, ECE

Abstract: Knowledge of the state of protein levels expressed in a cell, referred to
as the proteome, is expected to contain a wealth of information that may lead to
early diagnosis of diseases and better drugs. However, large-scale multiplexed
sensing of the proteome is a big challenge due to the large number of proteins
present and the various dynamic modifications they undergo. I will describe the
analytical techniques that are available today to undertake this task, particularly
focusing on the so called label-free bio-molecular sensing techniques. I will
describe the current state of the art of these techniques and speculate on ways to
address the current challenges in this field. Ongoing work in our group in this area
will be mentioned.

Title: Upper Bounds for Entropies of Sums, and the Ruzsa Divergence

Speaker: Prof. Mokshay Madiman, University of Delaware and Yale University

Date: 26 April 2013 (Friday)

Time: 4:00 pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Seminar Hall, ECE

Abstract:

Entropy inequalities for sums are ubiquitous in the study of fundamental limits of
communication systems. Much effort has gone into obtaining lower bounds on
entropies of sums (in both discrete and continuous settings, in each of which the
behaviour is quite different). Our focus will be upper bounds, where we find that we
can give a unified analysis of bounds for both settings. These bounds imply, for
instance, the (new and non-obvious) fact that for i.i.d. random variables X and X',
the entropies of X+X' and X-X' strongly constrain each other, even though the
difference of these two entropies can be arbitrarily large (as shown by Lapidoth and
Pete). More general bounds can be given without the i.i.d. assumption, but to
develop them in a natural way requires the introduction of the new notion of Ruzsa
divergence, whose properties we develop.

The work on the discrete setting is joint with Adam Marcus (Yale) and Prasad
Tetali (Georgia Tech), and that on the continuous setting is with Ioannis
Kontoyiannis (AUEB).

Title: Association in IEEE 802.11 WLANs

Speaker: Prof. Joy Kuri, Dept. of Electronic Systems Engineering

Date: 05 April 2013 (Friday)

Time: 4:00 pm (tea/coffee at 3:45pm)

Venue: Golden Jubilee Seminar Hall, ECE

Abstract:

Station-Access Point (STA-AP) association is an important function
in IEEE 802.11 Wireless LAN (WLAN) management. We obtain
an association policy that can be implemented in centralized WLAN
management devices, or in STAs, by taking into account explicitly two
aspects of practical importance:(a) TCP-controlled short file downloads
interspersed with read times (motivated by web browsing), and
(b) different STAs associated with an Access Point (AP) at possibly
different rates (depending on distance from the AP).

First, we consider an analytical model to obtain the aggregate AP
throughput for long TCP-controlled file downloads when STAs
are associated at k different rates r1,r2,...,rk; this extends
earlier work in the literature. Second, we present a 2-node closed
queueing network model to approximate the expected average-
sized file download time for a user who shares the AP with other
users associated at a multiplicity of rates. These analytical results
motivate the proposed association policy, called the Estimated Delay
based Association (EDA) policy: Associate with the AP at which the
expected file download time is the least.

Simulations indicate that for a web-browsing type traffic scenario,
EDA performs substantially better than other policies that have been
proposed earlier. Crucially, the improved performance is sustained
even in realistic evaluation scenarios, where the assumptions
underpinning the analytical model do not hold. To the best of our
knowledge, this is the first work that proposes an association policy
tailored specifically for web browsing. Apart from this, our analytical
results could be of independent interest.

Title:  Can We Understand Biological Vision From an Engineering Perspective?

Speaker:  Prof. S.P Arun,  Centre for Neuroscience and ECE Dept., IISc

Date:  22nd February 2013 (Friday)

Time:  4:00 pm    (tea/coffee at 3:45pm)

Venue:  Golden Jubilee Seminar Hall, ECE

Abstract:
 

=======

With the advent of powerful computers and neural networks in the 1970s it
was thought that it would be extremely easy to make a computer see like us.
But today it is increasingly clear that vision is an extremely challenging
computational problem that remains still unsolved by computers today. In
the purest sense it is ill-posed because it involves reconstructing a 3-d world
from a 2-d image. Yet our brains have managed to solve this problem over
the course of evolution. How does our brain pull off such efficient vision?
Will engineering concepts and techniques be useful in understanding biological
vision? In my talk I will introduce this problem and outline how research in my
lab is attempting to answer specific aspects of these general questions.
 

The electronics Integrated Circuit (IC) chips have enabled us to create intelligence
on silicon (Si). Hence the impact of electronics manufacturing cuts across the entire
industrial ecosystem. SIlicon Complementary Metal Oxide Semiconductor (CMOS) technology
is the enabling technology for high performance Silicon ICs.  The phenomenal growth in
the CMOS technology over the past few decades has enabled very high performance compute
and storage systems, powering the information technology revolution. The transistor has
 meta-morphed from dimensions of a few tens of microns from 1950s to a few tens of
nanometers today.  We have 32 nm CMOS technology in volume production enabling GHz
microprocessors and Gbit memories. Moore's law has continued to work, albeit with
several technological innovations. The underlying premise of this growth is the ability
to scale down the transistor dimension by 30%, in every two year cycle on an average.
The obvious questions raised in the recent past are – How far can we continue to
miniaturize transistors? Do we need different transistor structures in future? Is it
possible for the new semiconducting materials to replace silicon? How long can we push
Moore’s law? Do we need radically different information storage and processing constructs?
We will attempt to answer some of these questions, in this talk, providing contextual
reference to work done in our group and elsewhere.