Digital Modulation Recognition Method Based on Self-Organizing Map Neural Networks

In this paper, a novel automatic modulation recognition method is proposed based on unsupervised learning neural networks, which adopts self-organizing map (SOM) algorithm. Firstly the basic structure of the SOM neural networks is described, and then the traditional training algorithm of SOM is improved in terms of learning efficiency and classification performance. Practical communication signal is made use of testing the method performance. Experiment results illustrate that SOM neural networks based digital modulation identification method has higher recognition rate than the method based on back-propagation neural networks. The method proposed in this paper provides a new approach for communication signal modulation identification.