Open AccessMulti-modulation Recognition Using Convolution Gated Recurrent Unit Networks
Author(s) -
Jun Tan,
Limin Zhang,
Zhaogen Zhong,
Wenlong Yu
Publication year - 2019
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1284/1/012052
Subject(s) - softmax function , modulation (music) , computer science , convolutional neural network , pattern recognition (psychology) , convolution (computer science) , artificial intelligence , quadrature amplitude modulation , signal (programming language) , noise (video) , computational complexity theory , speech recognition , channel (broadcasting) , artificial neural network , algorithm , telecommunications , image (mathematics) , bit error rate , physics , acoustics , programming language
To solve the problems of long modulation and recognition time and complexity of convolutional neural networks, a modulation method based on CNN-GRU (Convolutional Neural Networks and Gated Recurrent Unit) is proposed. Firstly, the spatial features of the signal are extracted by CNN convolution operation, then the timing correlation of the signal is extracted by GRU and finally the recognition probability is output by using the softmax layer to achieve the purpose of multi-modulation recognition. The experimental results show that the recognition performance of the method is further improved under the condition of no prior information such as channel and noise, and 11 modulation categories such as 16QAM and 64QAM can be effectively identified, and the complexity of the method is low, which greatly saves. The training identification time has good engineering application value.