基于改进OpenMax的开集干扰识别方法

Open Set Interference Recognition Method Based on Improved OpenMax

摘要:

目前的无线通信干扰识别研究主要针对闭集场景,为了面对日益复杂的电磁环境,对无线通信开集场景下未知干扰信号识别的问题进行了研究,提出了一种基于改进OpenMax的开集干扰识别算法；首先将干扰信号从时频空间映射到特征空间进行聚类,采用了度量学习中基于距离的损失函数对网络进行训练,通过样本映射至特征空间的点到中心锚点的欧几里得距离判断其所属干扰类别或是未知干扰,然后引入Open-Max算法对样本边界进行拟合,实现对阈值的自适应调节,最后通过仿真实验对算法性能进行验证；实验结果表明,该算法对干扰信号的聚类效果优秀,可以在对已知干扰识别准确率超过90%的同时识别出未知干扰,更加适合动态环境下的干扰识别。

Abstract:

The current wireless communication interference identification research mainly focuses on closed set scenarios, in order to face the increasingly complex electromagnetic environment, the problem of identifying unknown interference signals in open set scenarios of wireless communication is investigated, and an open set interference identification algorithm based on the improved OpenMax is proposed; firstly, the interference signals are mapped from the time-frequency space to the feature space to be clustered, and the loss function based on the distance in the metric learning is used to train the network. The network is trained, and the Euclidean distance from the point of the sample mapped to the feature space to the center anchor point is used to determine whether it belongs to the interference category or unknown interference, and then the Open-Max algorithm is introduced to fit the sample boundaries to achieve adaptive adjustment of the threshold, and finally the algorithm is verified by the simulation experiments; the experimental results show that the algorithm has excellent clustering effect on the interference signal, and it can identify the known interference in more than 90% of the accuracy. The experimental results show that the algorithm has excellent clustering effect on the interference signal, and can recognize the unknown interference with an accuracy of more than 90% of the known interference, which is more suitable for the interference recognition in the dynamic environment.