基于粒子群支持向量机的通信信号调制识别算法

摘要

为了解决大部分通信信号调制识别方法计算量大和分类器训练困难问题,提出一种基于粒子群(PSO)支持向量机(SVM)的调制识别方法.将小波理论与调制信号的瞬时特征、高阶累积量以及分形理论相结合,得到一种混合模式特征向量,并利用粒子群支持向量机对2ASK,4ASK,2PSK,4PSK,8PSK,2FSK,4FSK,8FSK,16QAM和MSK 10种调制信号进行分类识别.仿真结果表明当信噪比大于等于5 dB时,信号正确识别率大于等于98％.%To solve the problems of most communication signals modulation recognition methodscomputational complexity and classifier training difficulties, a method of modulation recognition is proposed based on Particle Swarm Optimization(PSO) and Support Vector Machine(SVM). Combined wavelet theory with the modulated signals instantaneous characteristics, high-order cumulants and fractal theory to obtain an hybrid model of feature vector, and use PSO and SVM to identify ten kinds of modulation signals as 2ASK, 4ASK, 2PSK, 4PSK, 8PSK, 2FSK, 4FSK, 8FSK, 16QAM and MSK. The simulation results show that the success rate is over 98% when SNR over 5 dB.