基于SiameseNet网络深度学习的移动终端智能辨识方法研究

打开文本图片集

中图分类号：TN91 文献标志码：A

Researchofradioidentificationmethodbasedon SiameseNet deep learning model

ZHOU Shaoqiong，ZHU Xiaoyu，LIAO Shuai，XI Rui （Beijing Institute of Remote Sensing Information，Beijing lOOOl1，China）

Abstract： With the increasingly demand for communication and the continuous improvement of information technology，communication radio become more and more popular. The dificulty of monitoring and identifying radio is increasing while electromagnetic environment comes more sophisticated. Traditional radio identification method first extracts the expert features such as frequency and then uses the clasifier to identify.This method has many disadvantages such as high complexity，complete prior information and limited application range. The deep learning theory which proposed a method for radio identification based on SiameseNet Model is comprehensively used，and model training and testing in lab is carried out. The measured results show that the SiameseNet deep learning model can achieve the high recognition rate of 90% ：

Keywords：deep learning；SiameseNet model;radio fingerprint

0 引言

目前，市面上的移动终端由于生产厂家、生产工艺等不同，使得即使同一型号的终端发射的信号也存在独一无二的细微特征，这些细微特征通常包含载荷的硬件特性和平台的物理特性，能够作为信号发射个体的身份标识，因此也被称为移动终端的指纹，终端辨识也被称之为指纹识别[。（剩余4251字）