LDPC codes over famous graphs

Irina Bocharova

State University on Information Technologies, Mechanics and Optics, St.-Petersburg, Russia

Abstract:

An overview of graphical representations for low-density parity-check (LDPC) codes is given. A parallel between graph (hypergraph)-based codes and LDPC codes is drawn based on these representations. Constructions of generalized LDPC codes called woven graph codes with constituent block and convolutional codes are considered. Asymptotical bounds on the minimum (free) distance of woven graph codes are presented. Examples of woven graph codes with convolutional constituent codes having large and extremely large free distances are given. Simulation results of maximum-likelihood decoding of woven graph codes are presented and discussed. Relation to nonbinary LDPC codes is explained. Examples are given.

Biography:

Irina E. Bocharova was born in Leningrad, U.S.S.R., 1955. She received the Diploma in Electrical Engineering in 1978 from the Leningrad Electro-technical Institute and the Ph.D. degree in technical sciences in 1986 from Leningrad Institute of Aircraft Instrumentation.

Since 2007 she has been Associate Professor at the State University of Information Technologies, Mechanics and Optics. Her research interests include convolutional codes, communication systems, source coding and its applications to speech, audio and image coding. She has more than 50 papers published in journals and proceedings of international conferences, and seven U.S. patents in speech, audio and video coding. She has authored the textbook «Compression for Multimedia», Cambridge University Press, 2010.

Professor Bocharova was awarded the Lise Meitner Visiting Chair in engineering at Lund University, Lund, Sweden twice (January--June 2005 and 2011). In 2014 she received Scopus Russia Award.