Syndrome-based neural decoding (SBND) has emerged as a promising deep learning approach for soft-decision decoding of high-rate, short-length codes. However, this approach still has substantial room for improvement. In this paper, we show how to leverage code automorphisms to enhance the ability of existing SBND models to learn and generalize through data augmentation during training and inference. As a result, for the short high-rate codes considered, we obtain models that closely approach MLD performance using small datasets and proper training. Our findings also suggest that many prior results for SBND models in the literature underestimate their true correction capability due to undertraining. Code to reproduce all results is available at: https://github.com/lebidan/sbnd.
Syndrome-based neural decoding (SBND) has emerged as a promising deep learning approach for soft-decision decoding of high-rate, short-length codes. However, this approach still has substantial room for improvement. In this paper, we show how to leverage code automorphisms to enhance the ability of existing SBND models to learn and generalize through data augmentation during training and inference. As a result, for the short high-rate codes considered, we obtain models that closely approach MLD performance using small datasets and proper training. Our findings also suggest that many prior results for SBND models in the literature underestimate their true correction capability due to undertraining. Code to reproduce all results is available at: https://github.com/lebidan/sbnd.
Recently, Machine Learning has been considered for use in various communications subsystems, providing an alternative paradigm for addressing core problems in the physical (PHY) and medium access control (MAC) layers. Several methods and neural network topologies have been proposed for channel decoding targeting short and moderate length codes. In this paper, we consider the Syndrome-based Deep Learning Decoder and propose a modification in order to enable iterative operation. The proposed modifications perform iterations of the original decoder, which are unrolled to build a larger network. The unrolled iterations are subsequently trained jointly. Moreover, the proposed iterative functionality is incorporated within the learning process by accounting for the error of each iteration in the loss function. By sharing the weights between the iterations, we show that it is possible to reduce the storage requirements of the original decoder by a factor of 2.6x for BCH(63,45), while approaching maximum-likelihood decoding performance. Finally, simulation results are provided for several code lengths and rates, which demonstrate the error-correcting capability of the proposed iterative decoder.
Error correction at short blocklengths remains a challenge for low-density parity-check (LDPC) codes, as belief propagation (BP) decoding is suboptimal compared to maximum-likelihood decoding (MLD). While BP rarely makes errors, it often fails to converge due to a small number of problematic, erroneous variable nodes (VNs). Multi-round BP (MRBP) decoding improves performance by identifying and perturbing these VNs, enabling BP to succeed in subsequent decoding attempts. However, existing heuristic approaches for VN identification may require a large number of decoding rounds to approach ML performance. In this work, we draw a connection between identifying candidate VNs to perturb in MRBP and estimating channel output errors, a problem previously addressed by syndrome-based neural decoders (SBND). Leveraging this insight, we propose an SBND-inspired neural network architecture that learns to predict which VNs MRBP needs to focus on. Experimental results demonstrate that the proposed learning approach outperforms expert rules from the literature, requiring fewer MRBP decoding attempts to reach near-MLD performance. This makes it a promising lead for improving the decoding of short LDPC codes.
In this work, we introduce a framework that enables the use of Syndrome-Based Neural Decoders (SBND) for high-order Bit-Interleaved Coded Modulations (BICM). To this end, we extend the previous results on SBND, for which the validity is limited to Binary Phase-Shift Keying (BPSK), by means of a theoretical channel modeling of the bit Log-Likelihood Ratio (bit-LLR) induced outputs. We implement the proposed SBND system for two polar codes (64, 32) and (128, 64), using a Recurrent Neural Network (RNN) and a Transformer-based architecture. Both implementations are compared in Bit Error Rate (BER) performance and computational complexity.
We present a novel framework for applying deep neural networks (DNN) to soft decoding of linear codes at arbitrary block lengths. Unlike other approaches, our framework allows unconstrained DNN design, enabling the free application of powerful designs that were developed in other contexts. Our method is robust to overfitting that inhibits many competing methods, which follows from the exponentially large number of codewords required for their training. We achieve this by transforming the channel output before feeding it to the network, extracting only the syndrome of the hard decisions and the channel output reliabilities. We prove analytically that this approach does not involve any intrinsic performance penalty, and guarantees the generalization of performance obtained during training. Our best results are obtained using a recurrent neural network (RNN) architecture combined with simple preprocessing by permutation. We provide simulation results that demonstrate performance that sometimes approaches that of the ordered statistics decoding (OSD) algorithm.
This paper proposes a newly developed layer-wise syndrome-based framework for decoding Low-Density Parity-Check (LDPC) codes using deep learning. The framework leverages a layer-specific training approach to design methodologically effective decoding strategies, aiming to utilize the specific characteristics of each layer and improve overall decoding performance. By integrating the Normalized Offset Min-Sum (NOMS) algorithm into a Forward-Feedback Neural Network (NOMS-FF), the proposed model employs a soft-syndrome loss function to assist the learning of codeword structures and optimize decoding performance across varying noise levels. Additionally, the framework incorporates a layer-freezing strategy and dynamic Signal-to-Noise Ratio (SNR) allocation, enabling targeted post-training adjustments tailored to layer-specific characteristics. Experimental results on 6G-candidate QC-LDPC codes (Base Graph 2, N=240,K=80 ) demonstrate that the proposed approach provides substantial performance advantages over traditional decoders across the entire SNR range tested in this study. Furthermore, this study conducts partial-layer decoding experiments, which demonstrate that using only a subset of neural network layers can reduce the model’s computational runtime while simultaneously maintaining or improving decoding accuracy compared to the full-layer model under specific noise conditions.
The decoding of error-correcting codes (ECCs) is a critical aspect of communication systems, yet traditional decoding techniques can often be computationally demanding or ineffective for certain codes, necessitating innovative approaches. In this study, we introduce a hybrid approach that combines machine learning and automorphism techniques to optimize the decoding process. Specifically, we train multilayer perceptron (MLP) models to learn the mapping between error syndromes and their corresponding errors. While these models exhibit robust learning capabilities, their performance sometimes does not reach 100%. To mitigate this limitation, we exploit the automorphism group of the code—a set of structure-preserving transformations—to convert the errors that the MLP struggles to decode into ones it can process more effectively. We use a minimum number of p permutations, pre-calculating and storing all possible automorphisms to ensure computational efficiency. Our experimental results reveal that this hybrid approach substantially enhances the decoding performance of the MLP model, presenting a promising avenue for decoding ECCs. Importantly, this approach is not limited to MLP models and can be applied to any machine learning model with a learning score less than 100%, broadening its applicability and impact. By integrating machine learning with traditional algebraic coding theory, we propose a new paradigm that holds the potential to revolutionize the design of decoding systems, making them more efficient and effective.