Automatic Modulation Recognition Based on Lightweight Network Under Alpha Stable Distributed Noise

‍ ‍Automatic modulation recognition has played a huge role in wireless communication. Most studies hypothesis additive Gaussian noise channels no longer meet the accurate description of the actual channel environment. In practice， Alpha stable distribution noise is widely presented due to lightning， thunderstorm， multi-user interference， equipment failure， etc. Therefore， it is more in line with the actual and challenging. This article proposes a modulation recognition method for preprocessing a combined lightweight network for Alpha stable distribution noise. First， by pretreatment of the receiving signal to the digital domain mapping and threshold limit， the sharp pulse caused by the Alpha stable distribution noise is controlled to a reasonable range； then， a lightweight network based on the Ghost module is proposed. Complete the signal’s modulation identification classification task. The experimental results show that the method mentioned in this paper has high recognition accuracy and lower computational complexity compared to existing CLDNN， CNN， ResNet.