Research On Methodology For Passive Multistatic Localization With Auxiliary Information

Passive multistatic localization is a fundamental technique which has been exten-sively studied and with a wide variety of applications in past decades.Depending on the signals radiating from the target or not,the problems of passive multistatic localization can be classified into passive multistatic source localization and passive multistatic lo-calization with illuminators of opportunity(IOs).In the scenario of passive multistatic source localization,the received signals are radiated from the target.In the passive mul-tistatic localization with IOs,signals from the IOs are reflected by the target.In two-step methods for passive multistatic localization,the random errors in positions of receivers and unknown measurement noise covariance would result in degradation on accuracy of localization.This dissertation mainly focuses on the problems of passive multistatic localization for moving target,passive multistatic localization with erroneous positions of receivers and passive multistatic localization with unknown measurement noise covariance.Be-sides,the localization methods based on auxiliary information are proposed for each prob-lem.The main contributions of this dissertation are summarized as:1.For moving target in passive multistatic source localization and passive multistatic localization with IOs,closed-form solutions based on error refinement are proposed for each case.The first order Taylor-series expansion of auxiliary variables with respect to target position and velocity are adopted as auxiliary information to refine the error in estimate obtained from weighted spherical-interpolation.Both methods achieve better performance on the estimate for position and velocity of target than the state-of-art closed-form methods.2.To improve the localization accuracy in TDOA-AOA hybrid localization with erroneous positions of receivers,a calibration source and its measurements are used as auxiliary information.Considering that the position of calibration source may be known precisely or imprecisely,two closed-form methods are proposed.The closed-form meth-ods both consist of two stages.In the first stage,the errors in positions of receivers are estimated from the auxiliary information and then refined.In the second stage,the tar-get is localized with the refined positions of receivers.Both methods are able to attain the Cramér Rao Lower Bounds(CRLB)accuracy under mild conditions.The proposed methods also perform better than the existing closed-form methods without calibration.In addition,the optimum placement of calibration source and strategy for suboptimal po-sition of calibration source under a special condition are derived from theoretical analysis of CRLB.3.To solve the problem of stationary target localization in passive multistatic local-ization with an IO under unknown measurement noise covariance,the multiple measure-ments with respect to same target are adopted as auxiliary information for localization.Firstly,the CRLBs of target position and measurement noise covariance are derived from the multiple TD measurements and localization model.Then,an alternately iterative op-timization strategy,which is based on the closed-form localization method with known measurement noise covariance and closed-form shrinkage estimator for covariance ma-trix with known center point,is proposed.Without the prior knowledge of measurement noise covariance,the proposed method achieves the same localization accuracy as the state-of-art method performing with exactly known measurement noise covariance.4.To solve the problem of multiple moving targets localization based on TDOA-FDOA with unknown measurement noise covariance,the multiple measurements col-lected at different moments are adopted as the auxiliary information for localization.Based on the multiple moving targets measurement model with auxiliary information,the CRLBs for positions and velocities of all targets and measurement noise covariance are derived from the likelihood function firstly.By considering the positions and velocities of all targets and measurement noise covariance as stochastic parameters,the problem of joint estimation is transformed to a maximum a posterior(MAP)problem.Then,an iterative approach based on the variational Bayes inference is proposed to jointly estimate the posi-tions and velocities of targets and measurement noise covariance.Under mild noise con-ditions,the proposed method achieves the CRLB accuracy in the estimation of positions and velocities of targets and measurement noise covariance.