Research On UAV Recognition Based On Doppler Features

In recent years,as a new sensing technology wireless sensing has been paid more and more attention.Wireless sensing technology obtains the status of a target by capturing its influence on the wireless signal.Wireless sensing technology is widely used in more and more fields because of its good non-contact sensing ability.At the same time,with the booming development of unmanned aerial vehicle(UAV)field,UAV is more and more widely used in society.The wide application of UAV brings great convenience to people’s life,but also has some negative effects.For example,lawbreakers may use UAV to invade personal privacy or even pose a threat to national security.Therefore,effective monitoring of UAV is an effective means to reduce these negative effects.In this thesis,we explore and study a new method for UAV recognition by extracting Doppler features from wireless signals using wireless sensing technology.The main research contents of this thesis are as follows:Firstly,because linear array radar can only provide range information and signal angle of arrival(Ao A)information,it cannot localize UAV in three-dimensional space through a single measurement.In order to solve this problem,this thesis proposes a new idea of obtaining UAV location information through multiple measurements by rotating radar.Each measurement can obtain the range information and the Ao A information under the current angle.Then,based on multiple measurements,the constraint relationship model of UAV location and each measurement in space can be built.Finally,the maximum likelihood estimation method can be used to solve the UAV location information.Secondly,because radar can only measure the Doppler component of UAV rotor in the direction of radar line of sight,when the same UAV hovers in different locations,radar will measure different Doppler features,resulting in different Doppler features of the same UAV in different locations,thus affecting the identification accuracy.Since there is a geometric relationship between the measured Doppler features and the position of UAV,this thesis proposes to use the position information of UAV to correct the Doppler features and get the real velocity characteristics of the rotor,so as to eliminate the influence of UAV position.In order to realize UAV recognition,features with significant distinguishing ability are extracted from the corrected Doppler features based on deep network.Finally,in order to evaluate the effectiveness of the above algorithm,a Frequency-Modulated Continuous Wave(FMCW)radar operating at 77 GHz is used to verify the above idea.In order to verify the localization algorithm,this thesis measured the UAV hovering at four locations,and the experimental results showed that the localization errors were all less than 1m.In order to verify the recognition algorithm,this thesis compares the recognition accuracy before and after Doppler feature correction.Experimental results show that the recognition accuracy is greatly improved after correction,and the recognition accuracy is higher than 95% under the condition of four kinds of UAV.