基于瞬态信号特征的无人机数传电台识别研究

Research on UAV data transmission radio identification based on transient signal feature

针对采用开源飞控的消费级无人机探测存在识别率不高的问题,文中提出一种利用无人机数传电台瞬态信号特征进行无人机分类与识别的方法。首先,利用USRP X310采集无人机数传设备无线信号,通过能量和幅度方差轨迹完成瞬态信号的粗定位和精对准,设计无人机数传电台瞬态信号的提取方法;其次,利用3级Haar小波提取瞬态信号的包络特征,提出一种基于特征贡献度的加权KNN优化算法,利用瞬态信号包络特征进行射频指纹识别性能研究,并分析K值参数对识别性能的影响;最后,将加权KNN优化算法识别性能与Matlab工具箱中的KNN算法进行对比测试。测试结果表明,加权KNN优化算法可以提高无人机数传设备的识别准确性,对无人机数传设备个体识别平均正确率可达到87.9%,类型识别平均正确率可达到92.8%。说明文中方法对采用开源飞控平台无人机进行探测识别具有有效性,能够弥补其他无人机探测方式的不足。

As the consumer UAV using open source flight control has low recognition rate in the detection,a method of UAV classification and recognition based on the transient signal features of UAV data transmission radio is proposed. The USRP X310 is used to collect wireless signals from UAV data transmission radio,the coarse positioning and fine alignment of transient signals are completed by means of energy and amplitude variance trajectory,and the extraction method of the transient signals of UAV data transmission radio is designed. The 3-level Haar wavelet is used to extract the envelope feature of transient signals,and a weighted KNN optimization algorithm based on feature contribution is proposed. The identification performance of UAV radio frequency fingerprint is studied by means of the envelope feature of transient signals,and the influence of the K value parameter on the identification performance is analyzed. The comparison test of recognition performance between the weighted KNN optimization algorithm and the KNN algorithm in Matlab toolbox is conducted. The testing results show that the weighted KNN optimization algorithm can improve the recognition accuracy of UAV data transmission devices,the average accuracy rate of the individual identification can reach 87.9%,and the average accuracy rate of the type identification can reach 92.8%. It can be seen that the method is effective for the detection and identification of UAV using open-source flight control platform,and can make up for the shortcomings of other UAV detection methods.