Performance Improvements and Hardness Results for LDPC

SPEAKER: Professor Priti Shankar, Indian Institute of ScienceAbstract: A random combinatorial construction for near-regular LDPC codes is proposed. Near-regularity implies that the left or right degree of a node varies by at most one from the average left or right degree respectively. Bounds on achievable girth are derived. Experimental results indicate that these codes perform somewhat better than the well known regular Progressive Edge Growth (PEG) codes. The notion of node credibility is defined and used to modify the sum-product decoding algorithm. Performance improvements in the waterfall region are observed using this modification. Finally it is shown that computing the stopping distance of LDPC codes is computationally intractable.Bio: Priti Shankar is in the Department of Computer Science and Automation at the Indian Institute of Science, where she is currently Professor. Her interests are in algorithmic coding theory, automata and computability, and compiler theory.Host: Professor P. Vijay Kumar, vijayk@usc.edu

Location: Hughes Aircraft Electrical Engineering Center (EEB) - 248

Audiences: Everyone Is Invited

Contact: Mayumi Thrasher