Abstract:Deep learning based decoding networks have shown significant improvement in decoding LDPC codes, but the neural decoders are limited by rate-matching operations such as puncturing or extending, thus needing to train multiple decoders with different code rates for a variety of channel conditions. In this correspondence, we propose a Multi-Task Learning based rate-compatible LDPC ecoding network, which utilizes the structure of raptor-like LDPC codes and can deal with multiple code rates. In the proposed network, different portions of parameters are activated to deal with distinct code rates, which leads to parameter sharing among tasks. Numerical experiments demonstrate the effectiveness of the proposed method. Training the specially designed network under multiple code rates makes the decoder compatible with multiple code rates without sacrificing frame error rate performance.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the limitations of existing deep - learning - based decoders in handling codewords with different code rates in low - density parity - check code (LDPC) decoding. Specifically, traditional neural network decoders need to train multiple models for different code rates, which not only increases the storage cost but also raises the training complexity. In addition, these decoders perform poorly when faced with code - rate - matching operations (such as puncturing or spreading), limiting their flexibility in practical applications, especially in resource - constrained application scenarios, such as the Internet of Things (IoT). To overcome these problems, the paper proposes a rate - compatible (RC) LDPC neural decoding network based on multi - task learning (MTL). This network utilizes an LDPC code structure similar to Raptor codes and can dynamically adjust the network structure at different code rates through a parameter - sharing mechanism, thereby achieving support for multiple code rates. This method not only reduces storage costs and training complexity but also maintains frame error rate (FER) performance comparable to that of traditional decoding algorithms. The main contributions of the paper include: 1. **Proposing a new rate - compatible neural decoding network**: This network can adapt to multiple code rates without the need to train a separate model for each code rate. 2. **Utilizing the parameter - sharing mechanism**: By activating different neurons at different code rates, parameter sharing is achieved, reducing the storage requirements. 3. **Experimentally verifying the effectiveness of the method**: Through experiments on 5G NR LDPC codes, it is proven that the proposed RC neural decoder outperforms traditional decoding algorithms at different code rates and has performance comparable to that of separately trained neural decoders. In conclusion, this paper aims to develop an efficient LDPC neural decoding network that can adapt to multiple code rates in order to reduce storage and training costs while maintaining good decoding performance.

Rate Compatible LDPC Neural Decoding Network: A Multi-Task Learning Approach

PDD-Based Decoder for LDPC Codes with Model-Driven Neural Networks

Rate-compatible LDPC-RS product codes based on raptor-like LDPC codes

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

A Task-Guided Normalized Min-Sum Decoding Network for LDPC Codes-Based DJSCC

Performance Evaluation of Channel Decoding with Deep Neural Networks.

Boosting Learning for LDPC Codes to Improve the Error-Floor Performance

A UnetNND-BP Architecture for Channel Decoding Under Correlated Noise

Neural Adjusted Min-Sum Decoding for LDPC Codes

A Neural Network-Based Compressive LDPC Decoder Design over Correlated Noise Channel

Multi-rate LDPC decoder implementation for china digital television terrestrial broadcasting standard

Deep Learning-Aided Trainable Projected Gradient Decoding for LDPC Codes

Section-Wise Revolving NBP-Like Decoders for QC-LDPC Codes

A Low Complexity Decoding Scheme for Raptor-Like LDPC Codes

A Graph-Neural-Network Decoder with MLP-based Processing Cells for Polar Codes

Multi-rate LDPC Decoder for the Chinese DTTB Standard

The New Decoding Method of Rate Compatible LDPC Codes Based on the IR-HARQ Schemes

A Model-Driven Quasi-ResNet Belief Propagation Neural Network Decoder for LDPC Codes.

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

Neural Self-Corrected Min-Sum Decoder for NR LDPC Codes

Boosted Neural Decoders: Achieving Extreme Reliability of LDPC Codes for 6G Networks