Deep learning-based automatic modulation recognition: Combination of CNN and LSTM neural network

With the deepening development of communication technology, the technology of automatic modulation and recognition of communication signals has been more and more widely used in military and civilian fields. This paper mainly studies the implementation of automatic modulation recognition using deep learning as a computing tool, focusing on CNN neural network and LSTM neural network, and conducting simulation experiments on public data sets. Based on the original CNN neural network, this paper introduces the structure of LSTM neural network and combines the advantages of the two types of neural networks to explore a combined neural network that is superior to the originally used CNN network. The experimental results of this thesis show that introducing the features of dynamic time series modeling of LSTM networks into deep learning networks can capture the global and local information of signals more effectively and improve the accuracy of neural networks in automatic modulation recognition.