基于RBF神经网络的调制识别  

Modulation recognition of communication signals based on RBF neural networks

作　　者：刘澍[1] 王宏远[1] 

机构地区：[1]华中科技大学电信系,湖北武汉430074

出　　处：《系统工程与电子技术》2008年第7期1241-1244,共4页Systems Engineering and Electronics

基　　金：国家自然科学基金资助课题(60475024)

摘　　要：针对通信信号这种非稳定的、信噪比(SNR)变化范围较大的信号,利用遗传算法训练的径向基神经网络分类器对各种调制信号的特征矢量进行分类识别,充分发挥径向基神经网络的广泛映射能力和遗传算法的全局收敛能力,并在遗传算法中加入了梯度下降算子,克服遗传算法收敛速度慢的缺点,加快了遗传算法训练神经网络的速度,使得分类器的识别率和鲁棒性得到明显改善。仿真实验的结果证明了此方法的有效性和可行性。This paper mainly proposes an algorithm with which to implement the optimal classifier of RBF neural networks with genetic algorithms and to classify the modulation types of communication signals automatically. The method is according to the purpose of classification, using the advantages of both the non-linearity and adaptiveness of RBF neural networks, and combining with the algorithms of good classifier performance and robustness. Genetic algorithm has the characteristics of global convergence and heuristic learning with disadvantage of slow convergence rate. The gradient operator is used to the genetic algorithm for speeding up the convergence rate. It overcomes the drawbacks of the general classifier of neural networks. Test shows that the propose method has good performance.

关 键 词：遗传算法 径向基神经网络 特征矢量 调制识别 

分 类 号：TN911.3[电子电信—通信与信息系统] TP183[电子电信—信息与通信工程]