基于t-SNE算法的ABPSK信号个体识别  

ABPSK signal individual recognition based on t-SNE algorithm

作　　者：姚舜禹 王雪[1,2,3] 邹德财 李优阳[1,2,3] YAO Shun-yu;WANG Xue;ZOU De-cai;LI You-yang(National Time Service Center,Chinese Academy of Sciences,Xi'an 710600,China;University of Chinese Academy of Sciences,Beijing 100049,China;Key Laboratory of Precise Navigation and Timing Technology,Chinese Academy of Sciences,Xi'an 710600,China)

机构地区：[1]中国科学院国家授时中心,西安710600 [2]中国科学院大学,北京100049 [3]中国科学院精密导航定位与定时技术重点实验室,西安710600

出　　处：《时间频率学报》2019年第4期336-344,共9页Journal of Time and Frequency

基　　金：中国科学院青年创新促进会人才资助项目

摘　　要：同一通信体系下的ABPSK(aeronautical binary phase shift keying)信号都有着相同的前导码,传统信号识别方法无法通过相同的前导码部分准确地识别出信号源,且常用信号特征属于高维特征,非常容易引发维度灾难。采用前导码相同的ABPSK实际信号采集数据的前导码,使用t-SNE算法对实际采集信号的前导码以及双谱进行降维,并且把降维后信号单一特征输入分类器中,不仅有效地利用了信号数据的流形信息,而且显著提升了基于信号单一特征进行信号个体识别的正确率。The ABPSK(aeronautical binary phase shift keying) signals in the same communication systems have the same preamble. The conventional signal recognition methods cannot identify the signal source by the same signal preamble portion accurately. The common signal features are high-dimensional features, and lead to curse of dimensionality easily. Based on the collected ABPSK data with the same preamble, this study used the t-SNE algorithm to reduce the dimension of the signal’s preamble and the bispectrum, and the single feature after dimension reduction was also inputted into the classifier, thus it not only effectively used the manifold information of signal data, but also significantly improved the correct rate of signal individual identification by single signal feature.

关 键 词：t-SNE算法流形降维 信号个体识别 维度灾难 信号细微特征 

分 类 号：P127.1[天文地球—天体测量]