认知无线电中调制识别算法研究  

Study on Modulation Recognition for Cognitive Radio

作　　者：敖仙丹[1] 何世彪[1] 吴乐华[1] 

机构地区：[1]重庆通信学院

出　　处：《通信对抗》2008年第2期29-32,共4页Communication Countermeasures

基　　金：重庆市自然科学基金重点项目(2007ba2017)

摘　　要：通信信号的调制类型识别对于认知无线电这种智能通信系统具有重要研究意义。利用调制信号的循环谱相关特征，提取了5个特征参数，给出了各个参数随信噪比变化的曲线图。分类器基于RBF神经网络，采用“一类一个网络”结构，并从提高网络识别性能出发，构建了大容量和高质量的网络训练样本，能够扩大识别范围，提高识别精度。基于谱相关特征参数和神经网络分类器的算法能动态识别信号的调制方式，仿真结果验证了该算法在低信噪比下的有效性。Communication signals＇ modulation recognition has great signality for Cognitive Radio（CR） which is an intelligentized communication system. The five character parameters reflecting the differences of modulations were extracted exploiting the propriety of the cyclic spectral correlation features. Then, the graphs of each parameter changing with SNR were presented. The ＂one class one net＂framework based on RBF neural network was used as a classifier. According to recognition performance, the training swatch with large capacity and high quality was established so as to expand the range of recognition and improve the recognition precision. The modulation types can be identified dynamically based on spectral correlation character parameters and neural network classifier. Simulation result proves the validity of the technique in low SNR environment.

关 键 词：认知无线电 调制识别 谱相关 神经网络 

分 类 号：TN014[电子电信—物理电子学]