Research Of Multi-rotor UAV Detection Techniques Based On LTE Passive Radar

Nowadays multi-rotor unmanned aerial vehicles(UAV)play a more and more important role in the area of public safety,medical rescue,weather surveillance,filming,aerial remote sensing,live broadcast,etc.However in the meanwhile,the industrial applications of UAV may trigger potential dangers to some extent,which not only pose severe threats upon low altitude flight channels and personal privacy,but also can probably be manipulated by criminals to destruct important public facilities.Therefore,it has academic significance and practical values to investigate detection techniques for multi-rotor UAVs.Radar system is a promising solution to detect the multi-rotor UAVs because of its all-weather day or night operation ability.Specially,passive radar exploits the existing illuminators of opportunity(IOO)in the environment for target surveillance purposes rather than via emitting radio signals and receiving the reflected echoes.In this sense it is better for passive radar to detect the multi-rotor UAV with “low small and slow” characteristic in urban area.After more than 80 year development,passive radar has evolved to the second generation coherent detection based system,and is on the way to the third generation knowledge-aided system which detects the target by exploiting the environment and target signature.Therefore,this thesis is aimed to utilize the most advanced and popular Long Term Evolution(LTE)illuminator to detect the multi-rotor UAV aided by its radar target signature.Among the various IOOs in the urban area,LTE signal is preferred due to the massive distribution of 4G base station,benefits of broadband OFDM modulation,and good compatibility to the 5G mobile communication in the near future.Hence,this thesis is aimed to utilize LTE signals to detect multi-rotor UAV,which primarily involves the four parts as follows:Firstly,we investigate the radar target signature of multi-rotor UAV under the illumination of LTE signals and then rebuild the mathematical model of echo signal influenced by varying wind speed.(1)We use Computer Simulation Technology(CST)software to obtain the bistatic Radar Cross Section(RCS)characteristics of DJI Phantom 3 drone with and without rotors,and conclude that:(a)the bistatic RCS of multi-rotor UAV varies from-60 dBsm to-15 dBsm symmetrically according to the azimuth and pitch angles;(b)the contribution of rotors to the radar scattering characteristic of multi-rotor UAV is negligible.(2)We investigate the motion characteristics of drones in terms of the speed variation induced by varying wind speed and conclude that fastly-varying wind gust is the primary factor compared with wind shear and turbulence.(3)We set up the mathematical model of varying Doppler frequency in terms of oscillation signal,and then reconstruct the target echo signal model.Secondly,we present the feasibility and improvement investigations of LTE as the new IOO for passive radar.(1)We analyse the detection range of LTE,and conclude that:LTE can detect the multi-rotor UAV within the 800 m range area under the condition of10 ms coherent integration time.(2)We analyse the resolution ability,and conclude that:the range and speed resolution of LTE is the best among all the IOOs in the urban area,which can be used to detect the multi-rotor UAV.(3)We derive and compare the average ambiguity function(AAF)of Orthogonal Frequency Division Multiplexing(OFDM)and Filterbank Based Multicarrier(FBMC)as the examples of multi-carrier modulation signals,and then propose an OFDM AAF improvement algorithm in Doppler frequency dimension by applying FBMC's prototype filter.Simulation results illustrate that the proposed algorithm effectively eliminates OFDM AAF's high Doppler side lobes with nearly no cost of Doppler bandwidth.Thirdly,LTE reference signal reconstruction methods are investigated under different signal to noise ratio(SNR).(1)In the high SNR scenario,based on the cascade model of blind channel estimation,demodulation and reconstruction for LTE reference signal reconstruction,the classical Lloyd-Max algorithm is introduced into OFDM reference signal reconstruction for the first time,and then a low complexity subcarrier correlation based Multi-Level Llyod-Max(SC-ML-LM)algorithm is proposed.The simulation results show that:(a)the blind channel estimation error performance is correlated with the subcarrier index difference.When the subcarrier index difference is 0,the minimum mean square error(MMSE)converges to least square(LS)lower bound;(b)the bit error rate(BER)performance of SC-ML-LM is nearly uncorrelated with the subcarrier index difference.When the subcarrier index difference is less than 5,the BER performance is approximately 0.002 under the 30 dB SNR condition,which means the good OFDM reference signal reconstruction accuracy.Therefore,SC-ML-LM is a feasible scheme for OFDM reference signal reconstruction in passive radar.(2)While in the low SNR scenario,owing to the unreliable performance of blind channel estimation,the reconstruction result of random data is impossible.So we turn to exploit the pilot signals rather than the data signals in LTE frame structure,and then propose a novel LTE reference signal reconstruction method by only utilizing pilot signals to enable the robust cross correlation detection performance in low SNR reference channel.Numerical results show that the detection performance of the proposed pilots-aided LTE reference signal reconstruction algorithm is superior to that of the noisy reference signal scheme when the SNR in the reference channel is lower than approximately-11 dB.Therefore,the proposed pilots-aided LTE reference signal reconstruction algorithm is a feasible solution in low SNR environment for passive radar applications.Finally,the multipath clutter suppression and multi-rotor UAV detection estimation methods are investigated.(1)We analyze the clutter rejection performance of Extensive Cancellation Algorithm on Carrier(ECA-C)under the condition of imperfect reference signal,and conclude that the reference signal reconstruction error will only result in the residual clutter component other than adversely affecting target echoes.And then we revisit the classical ECA-C in the novel perspective of channel estimation,so that a low complexity subcarrier correlation based ECA-C is proposed by utilizing OFDM subcarrier correlation characteristic.The simulation results show that when the subcarrier index difference is 1,SC-ECA-C can reduce the computational complexity of ECA-C significantly with no cost of clutter suppression performance.(2)Inspired by the integration of parameter estimation and target detection,we set up the Doppler frequency's state transfer model for the scenarios with and without varying gust respectively by exploiting the motion characteristics,and then propose an interacting multiple model kalman filter based Doppler frequency estimation algorithm.The simulation results show that the proposed algorithm can detect the drone's echo as well as estimates Doppler frequency when the SNR is 0dB.