Deep learning algorithm featuring continuous learning for modulation classifications in wireless networks

作     者：WU Nan SUN Yu WANG Xudong WU Nan;SUN Yu;WANG Xudong

作者机构：School of Information Science and TechnologyDalian Maritime UniversityDalian Liaoning 116000China 

出 版 物：《太赫兹科学与电子信息学报》 (Journal of Terahertz Science and Electronic Information Technology)

年 卷 期：2024年第22卷第2期

页      面：209-218页

学科分类：11[军事学] 080904[工学-电磁场与微波技术] 0809[工学-电子科学与技术（可授工学、理学学位）] 08[工学] 110503[军事学-军事通信学] 0810[工学-信息与通信工程] 1105[军事学-军队指挥学] 1104[军事学-战术学] 082601[工学-武器系统与运用工程] 081105[工学-导航、制导与控制] 0826[工学-兵器科学与技术] 081001[工学-通信与信息系统] 081002[工学-信号与信息处理] 0811[工学-控制科学与工程] 

主　　题：Deep Learning(DL) modulation classification continuous learning catastrophic forgetting cognitive radio communications 

摘      要：Although modulation classification based on deep neural network can achieve high Modulation Classification(MC)accuracies,catastrophic forgetting will occur when the neural network model continues to learn new *** this paper,we simulate the dynamic wireless communication environment and focus on breaking the learning paradigm of isolated automatic *** innovate a research algorithm for continuous automatic ***,a memory for storing representative old task modulation signals is built,which is employed to limit the gradient update direction of new tasks in the continuous learning stage to ensure that the loss of old tasks is also in a downward ***,in order to better simulate the dynamic wireless communication environment,we employ the mini-batch gradient algorithm which is more suitable for continuous ***,the signal in the memory can be replayed to further strengthen the characteristics of the old task signal in the *** results verify the effectiveness of the method.