Parameter Estimation In The Passive MIMO Radar

Passive MIMO radar is a radar system that utilizes the signals already exist in the environment to detect,estimate,and classify the target under the multiple-input multiple-output setup.It possesses the advantages of both passive radar and MIMO radar,which has the potential to improve estimation performance.Parameter estimation is one of the main functions of a radar system.In many practical applications,the accuracy of parameter estimation determines the overall performance of a radar system.Therefore,the radar community usually uses the estimation performance to evaluate a radar system.In order to measure the estimation performance,a comprehensive evaluation metric is needed,people usually use Cramer-Rao bound(CRB),the lower bound of the mean square error of any unbiased estimator,as an evaluation metric.This thesis studies the problem of parameter estimation in passive MIMO radar.We analyze how different knowledge about the transmitted signals affects the performance of the parameter estimation.The maximum likelihood(ML)estimation and CRB are presented for different scenarios.The main contents are as follows:First,this thesis introduces several classic structures of passive radar.With an emphasis on the passive MIMO radar,we develop a signal model which describes the propagation characteristics of the passive MIMO radar and the parameters in the signal model are introduced.Then,this thesis studies the parameter estimation for the case of known signal form,but deterministic unknown signal parameters.For example,the signals are known to be GSM signals,with deterministic unknown bits.Through mathematical derivation,the expressions of the ML estimation and CRB are obtained.Using MATLAB simulation,the influence of the direct path to the estimation performance is analyzed.We present a comparison between this and the active MIMO radar case,and the performance loss is discussed.Next,this thesis focuses on the case where the signals are deterministic unknown.Again,the expressions of the ML estimation and CRB are provided.Further,for the case where the signals have constant modulus,we derive the ML estimation and CRB,and analyze the estimation performance via numerical examples.Finally,this thesis considers the random signal case.Under this case,only stochastic properties of the transmitted signals are known.Both centralized and distributed processing are presented,and the corresponding ML estimators and CRBs are derived.Through simulation,we verify the correctness of the analytical results.A comparison among the all above cases are given and the effects of different knowledge levels about transmitted signals on parameter estimation are summarized.