Learned Decimation for Neural Belief Propagation Decoders
Paper i proceeding, 2021

We introduce a two-stage decimation process to improve the performance of neural belief propagation (NBP), recently introduced by Nachmani et al., for short low-density parity-check (LDPC) codes. In the first stage, we build a list by iterating between a conventional NBP decoder and guessing the least reliable bit. The second stage iterates between a conventional NBP decoder and learned decimation, where we use a neural network to decide the decimation value for each bit. For a (128,64) LDPC code, the proposed NBP with decimation outperforms NBP decoding by 0.75dB and performs within 1dB from maximum-likelihood decoding at a block error rate of 10-4.

Författare

Andreas Buchberger

Chalmers, Elektroteknik, Kommunikation, Antenner och Optiska Nätverk

Christian Häger

Chalmers, Elektroteknik, Kommunikation, Antenner och Optiska Nätverk

Henry D. Pfister

Duke University

Laurent Schmalen

Karlsruher Institut für Technologie (KIT)

Alexandre Graell I Amat

Chalmers, Elektroteknik, Kommunikation, Antenner och Optiska Nätverk

ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

15206149 (ISSN)

Vol. 2021-June 8273-8277

IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Invited paper
Toronto, Canada,

Pålitlig och säker kodad kantberäkning

Vetenskapsrådet (VR) (2020-03687), 2021-01-01 -- 2024-12-31.

Styrkeområden

Informations- och kommunikationsteknik

Ämneskategorier

Telekommunikation

Sannolikhetsteori och statistik

Signalbehandling

DOI

10.1109/ICASSP39728.2021.9414407

Mer information

Senast uppdaterat

2022-04-06