Collaborative Passive Localization Based On Multidimensional Information

The rapid development of new techniques such as GNSS,navigation,UAVs and autonomous driving has spurred research in localization theory and industrial-military applications.Passive localization is one of the positioning technologies that uses a receiving system to obtain the transmitted signals or scattered signals to measure the source locations rapidly,while the positioning system itself does not transmit any signal during the process.Passive localization is worthy studying and has enormous potential in the military applications with long detection range and strong security.Based on a national defense research,this article aims to study the passive localization theory and algorithms of collaborative passive localization based on multidimensional information,and gives theoretical and technical support to the location problems in irrational environments.Nowadays,passive localization can be divided into two-step approach and direct position determination(DPD)method.It is worthy noting that the existing works only focus on how to improve the accuracy of positioning estimation and overcome the non-ideal factors such as noise,model errors and the ionosphere virtual height errors.However,it does not attempt to mine the multi-dimensional information to improve the accuracy from the collaborative sources.Aiming at updating the localization algorithms for different scenarios,our research work is carried out from four aspects: constraints of distance and velocity correlation,parameters of the signal modulation,known calibration sources and observation of ionosphere virtual height.The main contents and innovations are summarized as follows:1.A source localization problem using constraints of distance and velocity correlation for multiple disjoint sources moving together is studied first.Considering the multiple disjoint sources localization with constraints on distance and velocity correlation,an estimator is established based TDOA and frequency difference of arrival(FDOA)measurements.The inequality constraints were transformed into an exponential function as a penalty term.The constrained CRLB is derived to evaluate the performance of the proposed method.To improve the accuracy using the constraints,a Lagrangian estimator was developed and augmented cost function was established to solve the problem based on Newton's method.Compared with the unconstrained positioning algorithm,the positioning accuracy and the robustness have been improved,and it can achieve the constrained CRLB.2.The problem of multi-station DPD method based on the known parameters of signal modulation is studied secondly.Aiming at the radiation source whose signal modulation method is known,a digital modulation signal model is established by using the time difference of arrival(TDOA),the corresponding Cramér-Rao lower bound(CRLB)are derived under different conditions of signal waveforms.A DPD method is proposed based on the idea of alternating iterations,and the proposed method can solve the mixed-integer optimization of the symbol sequence and source position.Compared with the existing method for the unknown waveform,the positioning accuracy has been significantly improved using the proposed method.3.A DPD method can overcome sensor gain and phase errors(SGPEs)is proposed with the use of known calibration sources.Active calibration is one of the key technologies of the array errors calibration but less used for DPD methods.Considering a multi-station positioning scene with angle of arrival(AOA)and TDOA,a series of known calibration sources is introduced to establish the SGPEs.The CRLB is derived to evaluate the performance of the proposed method.A DPD method is proposed based on maximum likelihood estimate criteria,then the source position and the SGPEs are estimated by alternating iterations.Compared with the existing twostep method,the positioning accuracy of proposed method is higher especially at low SNR,and the accuracy attains the associated CRLB.4.The problem of over-the-horizon localization based on the ionosphere virtual height observation error is studied finally.Consider the scenario of ionospheric reflection,a signal model is first established based on AOA and TDOA.The observation noise and the observation error of the ionosphere virtual height are considered at the same time.We also derive the CRLB to evaluate the performance.Maximum likelihood estimate criterion is used to design the optimization model of source position and ionosphere virtual height.An over-the-horizon DPD method is proposed,which is more accurate than the MUSIC-type method and can reach the associated CRLB.