Abstract:To alleviate the suboptimal performance of belief propagation (BP) decoding of short low-density parity-check (LDPC) codes, a plethora of improved decoding algorithms has been proposed over the last two decades. Many of these methods can be described using the same general framework, which we call ensemble decoding: A set of independent constituent decoders works in parallel on the received sequence, each proposing a codeword candidate. From this list, the maximum likelihood (ML) decision is designated as the decoder output. In this paper, we qualitatively and quantitatively compare different realizations of the ensemble decoder, namely multiple-bases belief propagation (MBBP), automorphism ensemble decoding (AED), scheduling ensemble decoding (SED), noise-aided ensemble decoding (NED) and saturated belief propagation (SBP). While all algorithms can provide gains over traditional BP decoding, ensemble methods that exploit the code structure, such as MBBP and AED, typically show greater performance improvements.

What problem does this paper attempt to address?

This paper attempts to solve the problem of poor performance of short - length low - density parity - check (LDPC) codes in channel decoding, especially the sub - optimal performance based on the Belief Propagation (BP) decoding algorithm. Specifically, the paper aims to improve the decoding performance of short - length LDPC codes by comparing different integrated decoding methods, so as to achieve an effect close to maximum - likelihood (ML) decoding while maintaining low latency. ### Problem Background Short - length LDPC codes have important applications in the ultra - reliable low - latency communication (HRLLC) scenario, but their BP decoding performance is affected by the inevitable short cycles in the Tanner graph, resulting in performance degradation. To alleviate this problem, a variety of improved decoding algorithms have been proposed in the past two decades, and these algorithms can be classified into an integrated decoding framework. Integrated decoding processes the received sequence in parallel by multiple independent sub - decoders. Each sub - decoder proposes a candidate codeword, and finally, the maximum - likelihood (ML) decision is selected from these candidates as the output. ### Paper Objectives The objectives of the paper are to qualitatively and quantitatively compare several integrated decoding methods, including: - Multiple - Bases Belief Propagation (MBBP) - Automorphism Ensemble Decoding (AED) - Scheduling Ensemble Decoding (SED) - Noise - Aided Ensemble Decoding (NED) - Saturated Belief Propagation (SBP) These methods can all improve the performance of traditional BP decoding to varying degrees. In particular, those methods that utilize the code structure (such as MBBP and AED) usually show greater performance improvements. ### Main Contributions 1. **Comprehensive Review**: A detailed introduction to the integrated decoding methods and their adaptability to BP decoding. 2. **Mechanism Explanation**: An intuitive explanation of the construction mechanisms of different integrated decoding methods is provided. 3. **Performance Comparison**: Using Monte - Carlo simulations, these algorithms are quantitatively compared in terms of bit - error - rate performance, and the integrated recovery probability is introduced as an evaluation metric. Through these works, the paper shows how to significantly improve the decoding performance of short - length LDPC codes through appropriate code design and decoding strategies.

A Comparative Study of Ensemble Decoding Methods for Short Length LDPC Codes

New Low Complexity Belief Propagation Decoding of Low Density Parity Check Codes

High Performance Full Diversity LDPC Codes Construction Algorithm

Efficient Near Maximum-Likelihood Reliability-Based Decoding for Short LDPC Codes

Generalized reliability-based syndrome decoding for LDPC codes

Decoding Short LDPC Codes via BP-RNN Diversity and Reliability-Based Post-Processing

Reduced-Complexity Decoding of LDPC Codes

RELDEC: Reinforcement Learning-Based Decoding of Moderate Length LDPC Codes

Enhanced Quasi-Maximum Likelihood Decoding of Short LDPC Codes based on Saturation

Deep learning based enhancement of ordered statistics decoding of short LDPC codes

An Improved Decoding Algorithm for Finite-Geometry LDPC Codes.

Study of Weighted Residual Layered Belief Propagation for Decoding of LDPC Codes

Improved Generalized Automorphism Belief Propagation Decoding

The impact when neural min-sum variant meets ordered statistics decoding of LDPC codes

Locally-Optimized Reweighted Belief Propagation for Decoding LDPC Codes with Finite-Length

Spiking Neural Belief Propagation Decoder for Short Block Length LDPC Codes

Check Belief Propagation Decoding of LDPC Codes

Sparse Decoding of Low Density Parity Check Codes Using Margin Propagation

Boosting Ordered Statistics Decoding of Short LDPC Codes with Simple Neural Network Models

Neural Adjusted Min-Sum Decoding for LDPC Codes

Multistep Linear Programming Approaches for Decoding Low-Density Parity-Check Codes