Research On The Key Techniques Of Passive Localization

Passive location is a detecting technique for non-cooperative targets.It has strong concealment but not occupy the radio resource.It’s able to detect the small and over horizon targets.This dissertation focuses on the TDOA parameter estimation and multi-target trajectory tracking technology for passive detecting and locating consumer drones.And some algorithms are proposed for these complex practical scenarios.Relevant algorithms can also be applied to other applications such as radar or communication.In practical applications,the target’s signal may be composed of multiple subcarriers with unknown frequency point,whose amplitude spectrum is discontinuous and mixing with harmonic,leakage and interferences.In this dissertation,a PRANSCA is proposed to preprocess signals for such condition.The algorithm consists of two recursions.The first recursion aims to remove the interference and judge whether the signal exists through the amplitude spectrum.The second recursion filters out the phase outliers through random sampling method.Phase wrapping is also considered here,so as to become more practical.The experimental results show that the GCPS can work stably after preprocessing by the algorithm when it loses the estimation ability of TDOA with colored signal.The narrowband signal is commonly used by wireless targets.In processing the narrowband signal,GCPS algorithm only utilizes the phase gradient,so as to be susceptible to phase noise in narrowband.This dissertation proposes a JLS algorithm to improved the accurancy by utilizing both phase delay and group delay.JLS uses the phase gradient to estimate the phase wrapping parameter,and then calculates TDOA through a higher accurant phase-based method.However,the phase noise of non-Gaussian distribution impacts the estimation of phase wrapping.To further improve the accuracy,a PHAT-LSTM algorithm is proposed.The algorithm uses LSTM network as a nonlinear filter to extract the phase wrapping from the noise.Finally,a high-precision TDOA parameters is obtained by phase-delay algorithm.The simulation shows that,the RMSE of PHAT-LSTM under low SNR is significantly reduced.When the SNR is less than 10 d B,its RMSE is one-tenth of the GCPS methods.The buildings in the city will reflect the target’s radio wave.The multipath caused by these reflected path makes the GCPS algorithm unable to estimate TDOA.In order to solve the problem,the influence of the semi-open environment with reflection on the crosscorrelation phase spectrum is studied,and a DPTF model is constructed.DPTF model regards the model as a four-dimensional optimization problem.Compared with the already algorithms,DPTF algorithm needs no position information of the reflectors and reachs higher resolution.Based on DPTF,a Fast-DPTF is designed to accelerate it.Fast-DPTF makes use of the phase gradient to find the initial parameters,and then iteratively reduces the gradient fluctuation range to obtain more accurate values.This fast method can reduce the calculation time to 1/50 of the original algorithm while ensuring the effect.This algorithm is of great significance for expanding the deployment scene of observation array.MTT is a general technology in radar systems.However,most MTT algorithms face the problems of combination explosion and trajectory divergence.Especially in the application of passive source location,a large number of target and air collisions in the observation channel causing the problems more obvious.In this dissertation,a KF-SCAT algorithm is proposed,including a TDOA-based Kalman filter and a sequential clustering algorithm.KFs provide core points and neighborhoods for the clustering algorithm;clustering algorithm helps KFs filter out outliers and delete the invalid KF,so as to avoid trajectory divergence.In the outdoor test,when the classification accuracy of the traditional clustering algorithm for moving objects is only 31.2%,the accuracy of the KF-SCAT algorithm can reach 80.9%.The above algorithms are confirmed by a self-developed SEU-PSL system.A receiving array,a signal processing hardware platform,and a host computer are all part of the system.The hardware platform has 6 synchronous input chains to meet the needs of passive location.Its scanning frequency ranges from 70 MHz to 6.0 GHz covering most signal points.In order to meet the real-time requirements,the platform also accelerates some algorithms through the cooperation of software and hardware,which greatly compresses the data rate.To summarize,in terms of hardware platform design and signal processing algorithms,this dissertation focuses on the key technologies of the passive location.The proposed methods have been validated by both simulations and experiments and show practical significance for the progress and development of passive location technology.