Accelerating Pro-ADMM-HL Penalized Algorithm for Decoding Binary LDPC Codes

doi:10.1109/LCOMM.2024.3398596

Accelerating Pro-ADMM-HL Penalized Algorithm for Decoding Binary LDPC Codes

2024-07-01

1

PRE

AI

DOI：10.1109/LCOMM.2024.3398596

Original

Original request for help

Save

Abstract

En 中文

The alternating direction method of multipliers (ADMM) approach to decoding low-density parity-check (LDPC) codes has garnered lots of attention due to its superior error correction performance in error floor regions. However, improving convergence characteristic of ADMM decoding efficiently remains to be a crucial but challenging issue for practical application. In order to tackle this challenge, in this letter we first propose an efficient Pro-ADMM-based decoding algorithm framework by inserting additional quadratic proximal terms to each iteration of ADMM decoding for stabilizing the variable updates. Second, an message-passing interpretation of Pro-ADMM decoder is considered, and then an improved version based on horizontal layered scheduling, named Pro-ADMM-HL decoder is developed to further enhance the convergence behavior of Pro-ADMM decoding. Additionally, we analyze that the complexity of the proposed Pro-ADMM-HL decoders at each iteration displays linearly to LDPC code length. Finally, the effectiveness of the proposed Pro-ADMM and Pro-ADMM-HL decoding algorithm in terms of error correction performance and convergence characteristic is demonstrated by simulation results.

Keywords:

Decoding

Iterative decoding

Convergence

Scheduling

Error correction

Merging

Iterative algorithms

Low-density parity-check (LDPC)

alternating direction method of multipliers (ADMM)

horizontal layered scheduling

convergence characteristic

Accelerating Pro-ADMM-HL Penalized Algorithm for Decoding Binary LDPC Codes

Abstract

Journal

Organization