IEEE Home Search IEEE Shop Web Account Contact IEEE IEEE logo

IEEE Information Theory Society

Student Resources

 

Research Discussion and Lunch at CISS 2007

 
     

Home

Student Resources

Info. Theory Society

Media Services

 

Committee Events

Future Events

Past Events

 

Volunteering

 

Getting Started

Info. Theory Textbook
Info. Theory Tutorials

Interference Channel
Comm. Networks
Cooperative Comm.
Cryptography
Multiuser Info. Theory
Network Coding

Biology & Info. Theory
Quantum Info. Theory
 

Doctoral Dissertations

 

Finding Your Career

Job Resources

Job Opportunities

Pool of Candidates

 

Useful Links

Writing Papers

Journals

Conferences

Resources

 

Committee Members

 

The committee hosted a Student Roundtable Research Discussion and Lunch at the CISS conference at Johns Hopkins University on Wednesday, March 14. Approximately 60 students attended the lunch and enjoyed pizza and soda. Six discussion topics were chosen and each discussion was led by a student discussion leader. The discussion topics and leaders were:

Table #1:            Joint Source-Channel Coding for Wireless Networks

                           Leader: Deniz Gunduz (Polytechnic)

 

Shannon's source-channel separation theorem is one of the most fundamental results of information theory (its proof can be found in most information theory textbooks). It leads to the division of source and channel coding subproblems without losing the end-to-end performance. However, there are certain limitations to this result, which were the main discussion points of our table.

We first talked about the scenario of transmitting a Gaussian source over a slow-fading wireless MIMO channel where only the receiver has channel state information. In this scenario, the separation theorem fails due to the non-ergodicity of the channel. We talked about the existing results and possible extensions.

Next, we talked about multi-user networks, where there is no general source-channel separation theorem. We talked about certain special scenarios for which separation can be shown to be optimal, and others where separation is shown to be strictly suboptimal.

While it seems that a comprehensive tutorial on joint source-channel coding is lacking, and one including the most recent advances would be appreciated by the students of the topic, following report of an NSF workshop held in 1999 can be helpful to understand the fundamental problems and directions:

Workshop Report: NSF Sponsored Workshop on Joint Source-Channel Coding, San Diego, CA, Oct. 1999. (available at: http://www.ee2.caltech.edu/Faculty/effros/JSCC/index.html)


 

Table #2:            Convex Optimization Applications

                           Leader: Joydeep Acharya (Rutgers)

 

The topic at my table was convex optimization applications. Since we didn't have too many participants, we could engage in a one to one discussion about each person's research. We agreed that Prof Stephen Boyd's book is a good place to begin learning the subject. We then discussed about utility function based resource allocation problems in wireless settings. From there the discussion veered towards how interference leads to non convexity of the utility functions and possible ways to re-parameterize the objective and constraints to get rid of non-convexity.

For the beginner - Prof Mung Chiang's site is a good place to see the summary of research directions in convex optimization applications, results and open problems.

 

Table #3:            Detection and Estimation

                           Leader: Anima Anandkumar (Cornell)

 

Good reference books for the topic : "An introduction to signal detection and estimation " by Vincent Poor and "Statistical signal processing" by Steven Kay.
References for distributed detection and estimation:

  1. survey: Distributed detection with multiple sensors I. Fundamentals Viswanathan, R.; Varshney, P.K. Proceedings of the IEEE, Vol.85, Iss.1, Jan 1997 Pages:54-63

  2. asymptotic results: J. N. Tsitsiklis, "Decentralized detection by a large number of sensors,". Math. Control Signals Syst., vol. 1, no. 2, pp. 167182, 1988.

Table #4:            Distributed Source Coding

                           Leader: Peiyu Tan (Lehigh)

 

Distributed source coding (DSC) , also known as the Slepian-Wolf (SW) coding problem, concerns the separate encoding and joint decoding of two or more correlated sources. DSC is related to the CEO problem and is part of network information theory. The theoretical foundation was proposed by Slepian and Wolf more than 30 years ago, but the topics regarding practical applications of DSC become active recently, due to the boosting interests in sensor networks and wireless video.
One fascinating aspect of this problem is that, without knowing the instantaneous realization of the other source(s), the encoders can nevertheless achieve, by exploiting the statistics of source correlation, the same compression rate as joint encoding. Even more intriguing is that, although the problem is by nature a source coding one, the solutions lie in the technology of channel coding!

Milestones for distributed source coding:

  1. D. Slepian and J. K. Wolf, "Noiseless Coding of Correlated Information Sources," IEEE Trans. Inform. Theory, pp. 471-480, July, 1973.

  2. S. S. Pradhan and K. Ramchandram, "Distributed Source Coding Using Syndromes (DISCUS): Design and Construction," IEEE Tran. Inform. Theory, pp. 626-643, March, 2003.

Table #5:            Security and Cryptography

                           Leader: Alvaro Cardenas (Berkeley)

 

Information security is a very broad research field that usually focuses on providing services such as confidentiality, integrity and availability. The tools the information security researchers have developed to provide these services include cryptography, access control, intrusion detection etc. However with the need to provide new services such as anonymity, watermarking and traitor tracing, new adversary models (adaptive adversaries and bounded storage models) and new research paradigms such as quantum computation and quantum cryptography, the need to develop new techniques based on information theory, signal processing and game theory is evident.

In particular of interest to the information theory community is the new conference on information theoretic security sponsored by the International Association for Cryptologic Research (IACR). This year it is going to be held after EUROCRYPT in Spain:

http://www.escet.urjc.es/~matemati/maribel/ICITS/ITS07.htm


 

Table #5:            MIMO Systems

                           Leader: Mike Tinston (George Mason)

 

We had six people at the table and we all gave a brief overview of our work, discussed topics in MIMO systems, shared some of the references and who is doing what in the community, and ate a lot of pizza. Each of the participants described their research and we had an active exchange of information about the topic. We also recommended some references for our particular areas of research. We should have had someone jotting notes to remember what everybody said before we went our own ways and focused our attention back on our own problems.


Thanks to all of the discussion leaders for making the event a success!