基于CBAM-CNN-BiGRU的Morse信号智能识别译码算法研究

Research on Morse Signal Intelligent Recognition and Decoding Algorithm Based on CBAM-CNN-BiGRU

无线电Morse报是不可或缺的短波通信方式之一,但其自动接收译码方法研究较少,尤其是低信噪比环境下识别准确率、速度无法达到实用要求。提出了一种端到端的智能识别译码算法,对Morse信号进行时频变换和伪彩色图像增强;利用融入卷积注意力机制模块(Convolutional Block Attention Module,CBAM)的卷积神经网络(Convolutional Neural Network,CNN),完成有效特征序列提取;通过双向门控循环单元(Bi-directional Gated Recurrent Unit,BiGRU)实现识别译码及报文预测输出。图像增强与CBAM模块有效解决了小训练样本及低信噪比情况下,识别准确率过低的问题;BiGRU模块保证了译码实时性。算法针对仿真及实测数据的字识别准确率分别达到98.81%和96.68%,报文译码准确率分别达到96.76%和85.91%,明显优于同类算法。

Radio Morse signal is one of the indispensable short-wave communication methods,but its automatic receiving and decoding methods are less researched,especially in low SNR environment,the recognition accuracy and speed cannot meet the practical requirements.An end-to-end intelligent recognition decoding algorithm is proposed.First,time-frequency transformation and pseudo-color image enhancement are applied to Morse signal.Then,the Convolutional Neural Network(CNN)integrated with the Convolutional Block Attention Module(CBAM)is used to extract the effective feature sequences.Finally,the Bi-directional Gated Recurrent Unit(BiGRU)is used to realize identification and decoding and message prediction output.Image enhancement and CBAM modules effectively solve the problem of low recognition accuracy in the case of small training samples and low SNR.BiGRU module ensures real-time decoding.Based on the simulated and measured data,the character recognition accuracy of the algorithm can reach 98.81%and 96.68%respectively,and the message decoding accuracy can reach 96.76%and 85.91%respectively,which is obviously better than that of similar algorithms.