New Methods For Radar Network Performance Evaluation

This thesis considered multiple-input multiple-output(MIMO)radar network with widely separated antennas.Due to the use of multiple transmit and receive stations,MIMO radar has been shown to have advantages in detection and estimation.The performance of the joint estimation of target position and velocity using MIMO radar is studied.We investigate the impact of several non-ideal cases on the estimation performance and provide new approches to evaluate the performance.The Ziv-Zakai bound(ZZB)for the joint estimation is derived for the non-ideal cases.Unlike the CRB which cannot describe the estimation performance in the low signal-to-noise(SNR)region,the ZZB is a global bound and generally can predict the performance in the entire SNR region.Considering that the model mismatch is almost unavoidable in practice,the ZZB for the mismatched case is derived and the associated restictions is discussed.The asymptotic mean squared error(MSE)of the mismatch maximum likelihood(ML)estimator is provided,which can be used to evaluate the performance of the radar network under model mismatch.The thesis mainly contains the following parts:First,the principle of parameter estimation is introduced.The traditional signal model of MIMO radar for the ideal case is presented,based on which the Cramer-Rao bound for the joint estimation is discussed.Next,a more complicated but more realistic signal model for non-coherent MIMO radar is considered,which take into account possible non-ideal scenarios such as non-orthogonal signals,correlated reflection coefficients,and correlated noise components.This model has a broader area of applications,compared with the conventional one.Based on this model,the maximum likelihood estimates(MLE)and the corresponding ZZB for the joint estimation of target position and velocity are provided.The influence of signal orthogonality,the correlation of reflection coefficients,and the correlation of the noise components on the MIMO radar estimation performance is analyzed.Then,the mismatched signal model is developped.The estimation problem under model mismatch is studied and the mismatched MSE is discussed.Since the traditional ZZB cannot predict the value of the mismatched MSE,a mismatched ZZB is derived under certain constraints.Further,the asymptotic MSE of the mismatched estimator is derived.Finally,a MIMO radar network with mismatched model is considered.The asymptotic estimation performance of the radar network is analysed.It is shown that the previously derived asymptotic mismatched MSE can be employed as a tool to evaluate the performance of the radar network under model mismatch.