On the Design of LDPC-Convolutional Ensembles Using the TEP Decoder Articles uri icon

publication date

  • May 2012

start page

  • 726

end page

  • 729

issue

  • 5

volume

  • 16

international standard serial number (ISSN)

  • 1089-7798

electronic international standard serial number (EISSN)

  • 1558-2558

abstract

  • Low-density parity-check convolutional (LDPCC) codes asymptotically achieve channel capacity under belief propagation (BP) decoding. In this paper, we decode LDPCC codes using the Tree-Expectation Propagation (TEP) decoder, recently proposed as an alternative decoding method to the BP algorithm for the binary erasure channel (BEC). We show that, for LDPCC codes, the TEP decoder improves the BP solution with a comparable complexity or, alternatively, it allows using shorter codes to achieve similar error rates. We also propose a window-sliding scheme for the TEP decoder to reduce the decoding latency.

keywords

  • convolutional ldpc codes; finite-length analysis; tree-structured expectation propagation