| 摘要: |
| 针对当前通信信号调制识别算法在低信噪比环境下识别率不高、训练速度慢、识别类型受限等问题，提出了一种基于多特征麻雀搜索算法-极限学习机(Sparrow Search Algorithm-Extreme Learning Machine,SSA-ELM)的信号调制识别方法。该方法基于四阶和六阶累积量构造两个特征参数，引入分数阶小波变换，利用分数域小波系数构造特征值，组成三维特征向量后输入SSA-ELM网络进行分类。仿真及USRP(Universal Software Radio Peripheral)采集数据验证结果表明，所提特征参数具有较好鲁棒性，且SSA算法优化后的ELM网络分类性能得到明显提高，在SNR等于6 dB时两种方法识别率均达到90%，且最高识别率达到94%。 |
| 关键词: 通信信号 调制识别 分数阶小波变换 麻雀搜索算法 极限学习机 |
| DOI： |
|
| 基金项目:广西科技重大专项资助项目（桂科AA21077008）；桂林电子科技大学研究生教育创新计划项目（2020YCXS021） |
|
| Modulation recognition based on multi-feature SSA-ELM |
| XIAO Xiao,XIE Yuelei |
| (1.School of Information and Communication,Guilin University of Electronic Technology,Guilin 541004,China;1.School of Information and Communication,Guilin University of Electronic Technology,Guilin 541004,China;
2.Ministry of Education Key Lab. of Cognitive Radio and Information Processing(Guilin University of Electronic Technology),Guilin 541004,China) |
| Abstract: |
| In order to solve the problems of low recognition rate under low signal-to-noise ratio（SNR）,slow training speed and few recognition types of modulation in current modulation recognition algorithms,a modulation recognition algorithm based on multiple features and Sparrow Search Algorithm-Extreme Learning Machine(SSA-ELM) is proposed.The algorithm constructs two characteristic parameters based on fourth-order and sixth-order cumulants,and introduces a new time-frequency tool called Fractional Wavelet Transform,which uses fractional domain wavelet coefficients to construct eigenvalues and form three-dimensional eigenvectors,then uses the SSA-ELM network for classification.Verification result by simulation and Universal Software Radio Peripheral(USRP) collected data shows that the proposed feature parameters have good robustness,and the classification performance of the ELM network optimized by the SSA algorithm has been significantly improved.When the SNR is 6 dB,the recognition rate of both algorithms is 90%,and the highest recognition rate reaches 94%. |
| Key words: communication signal modulation recognition fractional wavelet transform sparrow search algorithm extreme learning machine |
