| 摘要: |
| 分析了一种物联网中元策略学习传输模式识别方法.采用阶段式元学习（ML）神经网络构成元学习器，利用数据增强（DA）技术对图像进行预处理，决策验证等多个模块被用于协作识别，能够有效地抵抗恶劣环境对信号造成的影响.理论分析和仿真结果证明了该调制模式识别方法的有效性. |
| 关键词: 自动调制识别 物联网 元学习(ML) 数据增强(DA) |
| DOI：10.3969/J.ISSN.1000-5137.2022.02.013 |
| 分类号:TP311 |
| 基金项目:浦东新区科技发展基金产学研专项(PKX2020-D6 |
|
| A meta-learning transfer pattern recognition method for Internet of Things |
|
XIU Sirui, ZHOU Xiaolin, WANG Baorui, DU Gang
|
|
School of Information Science and Technology, Fudan University, Shanghai 200433, China
|
| Abstract: |
| In this paper, a transfer pattern recognition method for meta-policy learning in the Internet of Things was analyzed. Firstly, the staged meta-learning (ML) neural network was used to construct a meta-learner. Secondly,the data augmentation (DA) technology was adopted to preprocess the image. Finally, multiple modules such as decision verification were used for collaborative recognition, which could effectively resist the influence of harsh environments on signals. The theoretical analysis and simulation results demonstrated the effectiveness of the modulation pattern recognition method. |
| Key words: automatic modulation recognition Internet of Things meta-learning (ML) data augmentation (DA) |
