Modulation Pattern Recognition of Communication Signals Based on Fractional Low-Order Choi-Williams Distribution and Convolutional Neural Network in Impulsive Noise Environment
Xiao-Di Tian, Xiaodong Sun, Xiaohui Yu, Xinbo Li · 2019
The majority of the existing automatic modulation recognition methods presume that the received wireless communication signal is corrupted by additive white Gaussian noise. The performance of the modulation recognition algorithms degrades severely under non-Gaussian impulsive noise. Hence, in this paper, we introduce a robust algorithm to identify the modulation type of radio signal contaminated by impulse noise by using convolutional neural network (CNN). The algorithm first employs fractional low-order Choi-Williams distribution (FLO-CWD) to get the time-frequency distribution of impulsive signal, then feature extraction and classification are performed with CNN, which has very high performance in image classification. Three types of digital modulation radio signals are used to validate the method. Simulation results demonstrate the superiority of the proposed method under impulse noise conditions.