Neural network improves error correction in data transmission codes
Learning to Decode Linear Codes with Generalized Belief Propagation Network
Artificial Intelligence
Summary
Sending data over noisy channels can cause mistakes, so error correcting codes help fix these errors. The authors present a new kind of neural network called GBPNet that improves how these codes are decoded by mimicking and extending traditional algorithms. Their method keeps the data structure intact while learning to better correct errors. Tests show that GBPNet reduces mistakes more than existing techniques.
What this means in practice
- •For communication hardware engineers: Improve decoding modules in communication devices by integrating GBPNet to reduce data errors transmitted over noisy channels.
- •For wireless system designers: Use GBPNet-based decoders in wireless systems to enhance reliability under challenging signal conditions without changing existing code structures.
Authors
Samira Malek
Abstract
Error correcting codes are a fundamental component of communication systems, enabling reliable data transmission over noisy channels. Recently, neural decoders have demonstrated notable performance gains over classical iterative algorithms. In this paper, we propose a Generalized Belief Propagation Network (GBPNet) that preserves message-passing symmetry while enhancing the decoder's learning capability. The architecture uses cascaded variable-node updates and learned element-wise message scaling while retaining the Tanner graph structure. We also introduce two training strategies, progressive training and high-SNR training, that improve decoding accuracy. Experimental results on several linear codes show that GBPNet achieves lower bit error rates than conventional belief propagation and the evaluated state-of-the-art neural decoders.