Modulation Recognition of Digital Signals Based on Deep Belief Network

2019 International Conference on Advanced Electronic Materials, Computers and Materials Engineering

无线电电子学;计算机科学;材料科学

Yunxin, Guo^1 ; Yue, Zhang^1 ; Hong, Ma^1

Department of Electronic and Optical Engineering, Space Engineering University, Beijing

101416, China^1

关键词: Deep belief networks;    Digital modulation signals;    Feature extraction and recognition;    High order cumulants;    Modulation patterns;    Modulation recognition;    Modulation signals;    Semi- supervised learning;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/563/5/052009/pdf
DOI  :  10.1088/1757-899X/563/5/052009

来源: IOP

【 摘 要 】

A modulation pattern recognition method for digital modulation signals, 4ASK, BPSK, QPSK, 2FSK and 4FSK digital modulation signals, which is based on deep learning model of deep belief network is proposed. The modulation signal is pre-processed and its high order cumulants are calculated as input training features. Solutions to the problem that the same high Modulation signals are generated in different SNR environments. Using the semi-supervised learning characteristics of deep confidence network, data sets are obtained to train the parameters of deep Confidence network layer by layer for feature extraction and recognition of modulation modes. The simulation results show that the recognition rate of this method is ideal.