重庆邮电大学学报（自然科学版）2025，Vol.37Issue(4)：500-506,7.DOI:10.3979/j.issn.1673-825X.202405230128
基于多判别器GAN的多终端干扰建模方法
Multi-terminal interference modeling method based on multi-discriminator GAN
摘要
Abstract
To effectively model complex interference in multi-terminal scenarios and address the challenge of multi-class da-ta generation,this paper proposes an improved multi-discriminator generative adversarial network(GAN)method.Each discriminator is trained to match a specific class of interference data distribution,thereby enhancing the generative perform-ance of the model.A communication link simulation environment is constructed to generate training data for interference sig-nals.The loss function of the network is analyzed,and the structure and training process of the GAN are detailed.Simula-tion results demonstrate that the proposed model outperforms the single-discriminator GAN in frequency-domain data fitting.The generated noise data closely match real measured data in terms of data distribution and average bit error rate(BER),with only a 0.1～2 dB difference in BER compared to actual datasets.关键词
多终端/生成对抗网络(GAN)/误比特率/干扰建模Key words
multi-terminal/generate adversarial network(GAN)/bit error rate/interference modeling分类
信息技术与安全科学引用本文复制引用
邓卢军,段红光,刘何鑫..基于多判别器GAN的多终端干扰建模方法[J].重庆邮电大学学报（自然科学版）,2025,37(4):500-506,7.基金项目
重庆市基础与前沿研究计划项目(cstc2019jcyj-msxmX0079) Chongqing Research Program of Basic Research and Frontier Technology(cstc2019jcyj-msxmX0079) （cstc2019jcyj-msxmX0079）
