Variable-Length Coding with Feedback: Finite-Length Analysis, Optimization and Code Construction
May 31, 2013
from 01:00 PM to 03:00 PM
|Where||Engr. IV Bldg., Maxwell Room 57-124|
|Contact Name||Tsung-Yi Chen|
|Add event to calendar||
Advisor: Richard D. Wesel
Increasing demand for high-quality, high-data-rate connections presents a significant challenge to system designers. Feedback helps to meet this challenge by allowing adaptation through communication from the receiver back to the transmitter. Commonly deployed feedback systems include automatic repeat request (ARQ) systems that ask for missed data packets to be retransmitted and hybrid ARQ systems that augment unsuccessful packet transmissions with additional error protection. This talk will show that the performance gap between these methods and the optimized variable-length coding scheme can be significant without careful system design. To demonstrate a system that operates with low latency and near-capacity throughput through an efficient use of feedback, this talk addresses the following: (1) analysis of feedback systems with various practical constraints, (2) optimization of the parameters related to an efficient feedback system, and (3) design of rate-compatible channel codes specifically for feedback systems. These foundations help to bridge the gap between feedback theory and practical implementation.
Tsung-Yi Chen is currently a Ph.D. candidate at the University of California, Los Angeles (UCLA), Department of Electrical Engineering. He is a recipient of the 2012-2013 UCLA Dissertation Year Fellowship. He received his M.S. degree from UCLA in 2009, and his B.S. degree from National Tsing Hua University in 2007, both in Electrical Engineering. His research is mainly focused on low-latency communication with feedback, which includes the analysis of variable-length feedback codes and the design of practical variable-length feedback codes.