基于改进RAN算法的调制分类器

资源分配网络（RAN）算法可以在线训练径向基（RBF）神经网络，但算法中的网络隐层中心通常从零开始，容易受噪声和异常数据影响。文章提出一种改进RAN算法，将遗传算法（GA）对数据优化聚类后的中心点作为RAN算法的初始中心。仿真表明，基于改进算法设计的RBF调制分类器，结构简洁，识别准确率高。

Modulation Classifier Based on Improved RAN Algorithm

Radial basis function neural networks can be trained with an online algorithm which is called resources allocation network. However, it starts from the empty structure and is often affected by the noise data. This paper proposes a new algorithm which initializes the starting centers properly by genetic algorithm. Simulation results show that the structure and the classification accuracy of the RBF neural network classifier are both improved.