Interference Management in Wireless Networks
Current wireless systems rely on a centralized interference management system
which tries to limit interference by centrally controlling medium access. As
wireless systems move towards a decentralized architecture (e.g.,
femto-cells), decentralized interference management is set to take
center-stage. However, we currently lack a fundamental theoretical
understanding of interference management in wireless systems. Some of my recent
research efforts have been directed towards addressing this gap.
-
``Interference Channels with Source Cooperation,'' V. Prabhakaran and P. Viswanath.
Submitted to IEEE Trans. Inform. Theory (pre-print)
-
``Interference Management Through Cooperation,'' V. Prabhakaran and P. Viswanath.
To appear at ISIT, 2009.
-
``Interference Channels with Source/Destination Cooperation,'' V. Prabhakaran and P. Viswanath.
Asilomar Conference on Signals, Systems and Computers, October, 2008. (paper)
We study the role of cooperation in interference management. By characterizing
the capacity to within a constant number of bits of a four-radio Gaussian
interference channels where the source/destinations may cooperate, we gain
insights into efficient modes of cooperation to achieve interference
management.
-
``Harnessing Bursty Interference,''
N. Khude, V. Prabhakaran, and P. Viswanath.
To appear ITW, 2009.(paper)
-
``Opportunistic Interference Management,''
N. Khude, V. Prabhakaran, and P. Viswanath.
To appear ISIT, 2009.
-
``Bounds on the Capacity Region of a class of Compound Interference Channels,''
A. Raja, V. Prabhakaran, and P. Viswanath.
ISIT, 2008. (Trans. on IT submission)
These papers investigate the issue of
limited channel state information (especially about the interfering links) that
may be available at the transmitter. We devise robust communication strategies
which allow the receivers to opportunistically decode larger amounts of
information depending on the state of interference. Combined with a layered
source code, these schemes provide graceful degradation of performance under
interference.
Relaying in Low-power Networks
In many low-power networks, the power cost for a node to remain ON to listen to transmissions from other nodes or to transmit to other nodes can constitute a significant part of the total power consumption by the radio. Thus, unlike the traditional relay channel model, under a low power constraint, the relay node cannot stay ON and listen to the entire duration of the transmission. We study low-power relay channels where the power cost of remaining ON is explicitly taken into account.
-
``Communication by Sleeping: Optimizing a Relay Channel Under Wake and
Transmit Power Costs,''
V. Prabhakaran and P.R. Kumar.
To appear at ISIT, 2009. (paper).
Secure Communication
It is well-known that stochastic resources (in the form of channels
and correlated sources) available at the communicating parties (and
potentially at eavesdroppers too) can be exploited to achieve secure
communication. My recent work has focused on utilizing these resources
available to the communicating parties in a point-to-point setting to
secure data from eavesdroppers.
-
``Secrecy via Sources and Channels -- A Secret Key - Secret Message Rate Trade-off Region,''
V. Prabhakaran, K. Eswaran and K. Ramchandran.
ISIT, 2008. (Extended arXiv draft).
-
``A Separation Result for Secure Communication,''
V. Prabhakaran and K. Ramchandran.
Allerton Conference, 2007. (paper)
We considered how different forms of stochastic resources
(correlated sources and channels) can be optimally combined to achieve secure
communication.
-
``On Secure Distributed Source Coding,''
V. Prabhakaran and K. Ramchandran.
IEEE Information Theory Workshop, Lake Tahoe, 2007. (IEEEXplore).
Motivated by sensor network scenarios, we
derived the optimal means of securing a source (sensor's observation) which
needs to be communicated over an open channel when that source itself is the
only form of stochastic resource available to the communicating parties to
achieve this.
-
``The Secrecy Capacity of a Class of Non-degraded Parallel Gaussian Compound
Wiretap Channels,'' T. Liu, V. Prabhakaran, and S. Vishwanath.
ISIT, 2008. (IEEEXplore).
The paper addressed the problem of secure communication over a wireless (MIMO)
channel in the presence of multiple eavesdroppers none of whom are degraded
with respect to the legitimate receiver. This is a departure from the results
currently available towards a more realistic setting, and uses a new coding
strategy which combines elements of network coding with more traditionally used
ideas.
Wireless Multicast
The considerable growth of wireless applications has brought to focus the
need for efficient use of the wireless spectrum. There is an immediate need
for technologies which can co-exist with currently deployed systems and
make efficient and opportunistic use of under-utilized or unused parts of
the spectrum. (Also see part II of my Ph.D. thesis.)
-
``Colored Gaussian source-channel broadcast for heterogeneous
(analog/digital) receivers,''
Vinod Prabhakaran, Rohit Puri, and Kannan Ramchandran.
IEEE Transactions on Information Theory,
(IEEEXplore)
This paper dealt with the transition of broadcast
radio and television from analog to digital technology. We derived a strategy
which achieves the optimal performance trade-off for digital and legacy analog
receivers using a hybrid analog-digital approach.
-
``Hybrid Digital-Analog Codes for Source-CHannel Broadcast of Gaussian Sources over Gaussian Channels,''
V. Prabhakaran, R. Puri, and K. Ramchandran.
(Trans. on IT submission)
-
``Hybrid Digital-Analog Strategies for Source-Channel Broadcast,''
V. Prabhakaran, R. Puri, and K. Ramchandran.
Allerton Conference, 2005. (conference paper)
In this paper, we tackled the problem of source-channel coding for
media-broadcast (better described by the term multicast in network information
theory circles) applications. We presented the currently best known strategy
for the long-standing open problem of source-channel broadcast of a Gaussian
source over Gaussian channels with bandwidth mismatch.
-
``MIMO Broadcasting with Common Information: A Deterministic Approach,''
V. Prabhakaran, S. Diggavi, and D. Tse.
Allerton Conference, 2007. (paper)
We also addressed the optimal broadcast strategy when a architecture which
respects source-channel separation is desired. The problem is that of degraded
message set broadcast which has long resisted efforts to solve for anything
more than a simple two-user scenario. We presented an exact solution for a
deterministic channel model which approximates a wireless MIMO channel.
Distributed source coding (for sensor networks)
The design of wireless sensor networks is well-known to present many
engineering challenges because of the limited capabilities of the sensors.
Part of my doctoral research involved deriving the fundamentals limits of
operation and optimal strategies for achieving these limits of simple
sensor networks performing the task of distributed estimation of a physical
process (e.g. temperature) from noisy observations made by a large number
of unreliable, resource-constrained sensors. We treated the problem as a
distributed source coding problem (known as the CEO problem). (Also see
part I of my Ph.D. thesis.)
-
``Rate region of the Quadratic Gaussian CEO Problem,''
Vinod Prabhakaran, David Tse, and Kannan Ramchandran.
ISIT, 2004. (Extended draft | IEEEXplore abstract)
-
``On the Role of Interaction Between Sensors in the CEO Problem,''
Vinod Prabhakaran, Kannan Ramchandran, and David Tse.
Allerton Conference, 2004. (paper)
Our main contributions
include (i) a characterization of the rate region of
the Gaussian CEO problem, (ii) optimality of the strategy under
sensor failuresand (iii) a characterization of the sum-rate of
the Gaussian CEO problem under sensor interaction and
feedback.
Distributed Networked Storage
A question of great interest in large-scale networks such as sensor
networks is how to enable ubiquitous access to data which gets generated
across many nodes in a distributed fashion.
-
``Decentralized Erasure Codes for Distributed Networked Storage,''
A. Dimakis, V. Prabhakaran, and K. Ramchandran.
Special Issue of IEEE Transactions on Information Theory: Networking
and Information Theory, July, 2006. (IEEEXplore)
The conference version received the best paper award at the IEEE/ACM Symposium on Information Processing in Sensor Networks (IPSN), 2005.
Using tools popularized by network coding, we looked at how data generated
across many sensor nodes can be stored by memory limited nodes such that a
data collector interested in accessing all the data may visit the least
number of nodes. The solution is in the form of a new distributed erasure
code.
-
``Distributed Fountain Codes for Networked Storage,''
A. Dimakis, V. Prabhakaran, and K. Ramchandran.
IEEE ICASSP, May 2006.
(IEEEXplore)
Here, we extended the formulation to the case where the
data collector tolerates the loss of a small linear fraction of the data.
Video Signal Processing
The video coders of today have achieved high levels of compression and
perform very satisfactorily under relatively clean channels. However, their
performance degrades considerably when the transmission suffers from fast
fades as in a wireless cellular network, especially under tight delay
constraints (e.g. video conferencing applications). This is because of
their reliance on the decoder having an exact copy of the previously
transmitted video frame. Distributed video coding is a radically different
approach which has better robustness properties using ideas borrowed from
multi-terminal source coding theory to video coding.
-
``Towards a Theory for Video Coding using Distributed Compression
Principles,'' P. Ishwar, V. Prabhakaran, and K. Ramchandran.
IEEE International Conference on Image Processing
(ICIP), 2003. (IEEEXplore)
We formulated and solved a joint remote-classification and
compression problem which models the distributed video coding system.
-
``Syndrome-based Robust Video Transmission over Networks with Bursty
Losses,'' J. Wang, V. Prabhakaran, and K. Ramchandran.
ICIP 2006. (IEEEXplore)
Following up on these innovations, we designed a novel scheme for transmission
of video in low-delay applications over channels with bursty losses.
Peer-to-peer Networking
-
``On the Role of Helpers in Peer-to-Peer File Download Systems: Design, Analysis and Simulation,'' J. Wang, C. Yeo, V. Prabhakaran, and K. Ramchandran.
International Workshop on Peer-to-Peer Systems (IPTPS). 2007 (Microsoft Research).
Peer-to-peer networks are becoming a very popular medium for content
distribution on the Internet. We proposed a strategy for improving the
efficiency of peer-to-peer networks by utilizing the spare capacity of idle
users in the system.
Last modified: Wednesday 6 May, 2009
