CESG Fishbowl (Room 333 WEB)
Prof. Sekhar Tatikonda
Dept. of Electrical Engineering
We study a new class of codes for Gaussian multi-terminal source and channel coding. These codes are designed using the statistical framework of sparse high-dimensional linear regression and are called Sparse Regression codes. Here codewords are sparse linear combinations of subsets of columns of a design matrix.
These codes have recently been shown to achieve the channel capacity of AWGN channels with computationally feasible decoding and to achieve the rate-distortion function for Gaussian sources with computational feasible encoding. Furthermore these codes can be used to implement random binning allowing us to develop coding schemes for a variety of Gaussian multi-terminal source and channel coding problems.
In this talk we first present sparse regression codes for both channel and lossy source coding; demonstrate how to implement random binning and superposition using these codes; and then show for a variety of multiterminal source and channel coding problems that these codes achieve the information theoretic limits and are computationally efficient. (Joint work with Ramji Venkataramanan, University of Cambridge.)
Sekhar Tatikonda received his Ph.D. in 2000 in Electrical Engineering and Computer Science from the Massachusetts Institute of Technology, Cambridge, MA. He was a postdoctoral fellow at the University of California, Berkeley from 2000-2002. In 2002, he joined the Electrical Engineering Department at Yale University, New Haven, CT, where he is currently an associate professor. His research interests are in communications, information theory, networked control systems, network optimization, and statistical machine learning.