Reduced-Complexity Linear Programming Decoding Based on ADMM for LDPC Codes

Haoyuan Wei

doi:10.1109/LCOMM.2015.2418261

2015-06-01

PRE

Haoyuan Wei *

Xiaopeng Jiao

Jianjun Mu

DOI：10.1109/LCOMM.2015.2418261

原文链接

原文求助

摘要

En 中文

The Euclidean projection onto check polytopes is the most time-consuming operation in the linear programming (LP) decoding based on alternating direction method of multipliers (ADMM) for low-density parity-check (LDPC) codes. In this letter, instead of reducing the complexity of Euclidean projection itself, we propose a new method to reduce the decoding complexity of ADMM-based LP decoder by decreasing the number of Euclidean projections. In particular, if all absolute values of the element-wise differences between the input vector of Euclidean projection in the current iteration and that in the previous iteration are less than a predefined value, then the Euclidean projection at the current iteration will be no longer performed. Simulation results show that the proposed decoder can still save roughly 20% decoding time even if both the over-relaxation and early termination techniques are used.

Keyword:

Linear programming decoding

alternating direction method of multipliers (ADMM)

reduce-complexity