Generalized ACE codes

SPEAKER: Mr. Aditya Ramamoorthy, EE Dept, UCLAABSTRACT: Low-Density Parity-Check Codes have been the subject of intense research lately because of their linear decoding complexity and near capacity achieving performance. Recent results show that LDPC codes drawn at random from irregular degree distribution ensembles can have significantly better performance asymptotically than regular LDPC codes and in fact provably achieve capacity over the binary erasure channel.At finite block lengths however one can achieve much better performance over randomly chosen codes by a careful construction of the parity check matrix. In this talk I shall explain the Approximate Cycle EMD (ACE) algorithm (Tian et al.) for the construction of LDPC codes for a given block length and degree distribution and discuss a generalization of their approach. I shall discuss the analysis of the expected stopping set spectrum of generalized ACE constrained ensembles which shows their improved performance over the binary erasure channel and the expansion properties of these codes that justifies their improved performance over the AWGN channel. Techniques for improving the performance even further at short block lengths will also be presented.This is joint work with Prof. Richard Wesel.BIO: Aditya Ramamoorthy is a Ph.D. candidate in the Electrical Engineering Department at the University of California, Los Angeles. He received his B. Tech. from the Indian Institute of Technology, Delhi in 1999 and his M.S. from UCLA in 2002. From 1999-2001 he was working as a systems engineer at Biomorphic VLSI Inc. In the summer of 2004 he was an intern at Microsoft Research working on network coding.His research interests include coding theory (in particular LDPC code design and analysis) and network coding.Host: Dr. Keith M. Chugg, x.07294, chugg@usc.edu

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

Audiences: Everyone Is Invited

Contact: Mayumi Thrasher