CRANet-Based Blind Recognition of Channel Coding

Blind channel coding recognition is highly useful in non-cooperative communication environments, such as cyber-electronic warfare. Blind channel coding recognition is a technique where the receiver identifies the channel coding scheme without prior knowledge of the coding or any additional data processing. In this paper, we propose CRANet, a network designed to improve channel coding recognition rates in blind environments by utilizing CNN, Residual, and Attention mechanisms. Eight types of channel coding schemes were used: BCH, Hamming, Product, RM, Polar, Golay, Convolutional, and Turbo. Simulation results show that the proposed CRANet outperforms benchmark deep learning models such as TextCNN and CNN-BLSTM, with accuracy improvements of up to 53.5% and 58.7%, respectively. Moreover, when using 2D CNN instead of 1D CNN, the recognition performance improved by 41.36% at –4 dB. Notably, CRANet with 2D CNN achieved an accuracy of 93.62% at 0dB.