摘要
提出了一种基于遗传算法与多层感知神经网络的调制识别方法,运用改进遗传算法优化的多层感知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.利用遗传算法的高效全局特性,克服了传统BP算法易于陷入局部最优解的缺点,同时在遗传算法基础上增加梯度下降算子,加快了收敛速度,使得分类器的识别率、收敛速度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.
A new approach based on genetic algorithm and MLP neural networks for the automatic modulation recognition of communications signals is presented. For the purpose of classification, we took the advantages of non-linearity and adaptiveness of MLP neural networks, combining with global convergence of the genetic algorithms. It overcomes the drawbacks of the general classifier of neural networks. local extremum and slow convergence speed. Requirements for a priori knowledge of the signals are minimized by the inclusion of an efficient carrier frequency estimator and low sensitivity to variations in the sampling epochs. Computer simulations indicate good performance on an AWGN channel, even at signal- to-noise ratios as low as 5 dB. This compares favorably with the performance obtained with most algorithms based on pattern recognition techniques.
出处
《武汉大学学报(理学版)》
CAS
CSCD
北大核心
2008年第1期104-108,共5页
Journal of Wuhan University:Natural Science Edition
基金
国家自然科学基金资助项目(60475024)
关键词
混合遗传算法
MLP神经网络
特征矢量
调制识别
hybrid genetic algorithm
MLP neural networks
characteristic vector
modulation recognition