基于GANomaly的水声通信信号调制方式开集识别
Open-set identification of modulation modes for underwater acoustic communication signals based on GANomaly networks
陈旗 1赵瑞轩 1唐劲松 1陈聪聪 2陆剑雄1
作者信息
- 1. 海军工程大学电子工程学院,湖北武汉 430000
- 2. 92866部队,山东青岛 266000
- 折叠
摘要
针对水声通信信号调制方式的开集识别问题,提出了一种基于GANomaly的新型开集识别方法.GANomaly网络模型结合了生成对抗网络(GAN)的生成能力和异常检测技术的判别能力,通过对抗训练,重构调制信号的时频特征并确定误差阈值,进而有效地区分已知和未知的调制方式.在信噪比>8 dB的条件下进行的仿真实验显示,该方法对已知调制方式信号的识别率均超过了86.00%,对未知调制方式信号的识别率超过了80.00%.仿真实验结果验证了所提方法对未知调制方式的有效识别能力,为水声通信领域开集识别问题提供了新的解决思路.
Abstract
For the open-set identification problem ofunderwater acoustic communication signal modulation modes,a new open-set identification method based on GANomaly was proposed.The GANomaly network model combines the generative ability of generative adversarial net-works(GAN)and the discriminative ability of anomaly detection technology.Through adver-sarial training,it reconstructs the time-frequency features of modulated signals and deter-mines error thresholds,thereby effectively distinguishing between known and unknown mod-ulation methods.Simulation experiments conducted under the condition of a signal-to-noise ratio greater than 8 dB show that the recognition rate of this method for signals with known modulation modes exceeds 86.00%,and for signals with unknown modulation modes,it exceeds 80.00%.The results of the simulation experiment verify the effective recognition ability of the proposed method for unknown modulation modes,providing a new solution for the open-set recognition problem in the field of underwater acoustic communication.
关键词
水声通信信号/开集识别/调制方式/生成对抗网络/时频特征Key words
underwater acoustic communication signals/open-set identification/modulation mode/GAN/time-frequency characteristics引用本文复制引用
出版年
2024
NETL
NSTL
万方数据