[Back]


Talks and Poster Presentations (without Proceedings-Entry):

A. Burg, A. Balatsoukas-Stimming, M. Meidlinger, G. Matz:
"Information optimized quantized message passing to enable ultra-high-speed (Tbps) LDPC decoding";
Talk: Information Theory and Applications Workshop (ITA 2016), San Diego (CA), USA; 01-31-2016 - 02-05-2016.



English abstract:
Polar codes are new and promising channel codes which are well-suited for numerous applications. Unfortunately, they have not yet been adopted by any communications standard, partly because it is unclear whether efficient hardware implementations of their decoding algorithms are possible. In this work, we attempt to provide some insight into this issue by presenting an efficient hardware architecture for successive cancellation list decoding. LDPC codes, on the other hand, are well-established and used in multiple recent communications standards. Standard message-passing decoding of LDPC codes is highly parallelizable, meaning that very high-throughput hardware implementations are possible. Using an information-theoretic quantization scheme to significantly reduce the message bit-width, we implement an LDPC decoder that can achieve a decoding throughpout of more than 1 Terabit/s.

Created from the Publication Database of the Vienna University of Technology.