Investigation On Target Detection And Direction Of Arrival Estimation For MIMO Radars

With the recent development in radar and multi-input multi-output(MIMO) communication techniques,MIMO radar has received vibrant attentions for radar engineers.The MIMO radars can be grouped into two classes:statistical MIMO(SMIMO) radar and co-located MIMO(CMIMO) radar.SMIMO radar is one with its transmitting and/or receiving elements separated far away from each other to achieve spatial diversity gain,while CMIMO radar applies waveform diversity to traditional phased array to form virtual aperture.In comparison with traditional radars,MIMO radars show performance advantages in target detection,parameter estimation,maneuvering target tracking,target identification and classification.On the basis of previous advances in MIMO radars,this dissertation investigates target detection and direction of arrival estimation(DOA) for MIMO radars.In the target detection,adaptive matched filter(AMF) based and image feature based target detection techniques are studied for airborne SMIMO radar in the presence of interference (including clutter and noise) with unknown statistical properties.Also disscussed here is the performance improvement by exploiting spatial diversity in SMIMO radar.In the DOA estimation studies,traditional DOA estimators are extended to SMIMO and CMIMO radars under symmetricα-stable(SαS) impulsive clutter environments.Performance improvements achieved by exploiting the inherent properties in MIMO radars are discussed.The Cramer-Rao Bounds(CRB) for estimating DOA of coherently distributed target(CDT) using CMIMO radar are finally studied.The main works and contributions are summarized as follows:1.Investigations on AMF based target detection technique for airbome SMIMO radar(1) The AMF detector for SMIMO radar is studied under interference with unknown statistical properties.Firstly,an SMIMO-AMF detector is proposed,and closed-form expressions of detection probability and false alarm probability are derived.Secondly,a simplified SMIMO-AMF detector is developed with the block diagonal property of interference covariance matrix to reduce the computational complexity,and a closed form expression of the detection performance is given for the special case of two receiving channels.The results show that the two detectors have constant false alarm rate(CFAR) with respect to the interference covariance matrix,and can effectively detect targets by exploiting the spatial diversity of SMIMO radar.(2) In both SMIMO-AMF detector and simplified SMIMO-AMF detector,it is required to compute the inverse matrix of the interference covariance matrix estimate.The independent and identically distributed(ⅡD) secondary data are often insufficient.The insufficient secondary data will cause the estimated interference covariance matrix being ill conditioned and therefore result in serious degradation in detection performance.This dissertation suggests the application of diagonal loading technique to SMIMO-AMF and simplified SMIMO-AMF detector.Two corresponding loaded detectors(SMIMO-LAMF and simplified SMIMO-LAMF detector) are developed and their performace are theoretically studied.Theoretical analyses indicate that the two loaded detectors have CFAR with respect to the interference covariance matrix.Simulation results show that the loaded detectors can improve the numerical condition of the estimated covariance matrix, and can improve detection performance by exploiting the spatial diversity of SMIMO2.Investigations on image feature based target detection technique for airborne SMIMO radarThe diagonal loading technique can improve the numerical condition of the estimated interference covariance matrix and the robustness of the AMF detectors.However,it can not reduce the requirement on the secondary data for covariance matrix estimation.This dissertation try to extend image feature space-time processing to airborne SMIMO radar and develop a new image feature based detector(SMIMO-IFSTP).The research shows that SMIMO-IFSTP detector can improve the detection performance by exploiting the spatial diversity to combat detection performance degradations induced by fluctuations of target radar cross section and small radial velocity of target,and can effectively detect ground moving targets without any requirement for interference covadance estimation by exploiting the distinct image features of targets and interference signals in the angle-Doppler domain.3.Investigations on DOA estimation for SMIMO radar and CMIMO radar in SαS impulsive clutterA lot of practical measurements have shown that radar clutter have impulsive characteristics,and can be well modeled as SαS distribution.The SαS distributions do not prossess finite second-order or higher moments.Consequently,many traditional DOA estimation algorithms can not be directly adapted to SαS impulsive clutters under stable law.To migitate the effects of SαS impulsive clutters,two pre-processing techniques, fractional lower order statistic(FLOS) and infinity-norm normalization,are suggested. With the pre-proceessed data,the tranditional DOA estimation algorithms can be used for MIMO radar to perform DOA estimations. (1) For SMIMO radar application,traditional Capon algorithm is applied after pre-processing.The FLOS-based Capon algorithm requires the clutter characteristic exponent value prior known or correctly estimated,whereas the infinity-norm normalization based Capon algorithm can work without these information.Computer simulations verify the effectiveness of the two pre-processing methods.The results also show that spatial diversity of SMIMO radar can improve the precision of DOA estimation in SαS impulsive clutter.(2) Most of the present DOA estimation algorithms for CMIMO radar assume the virtual sensor-spacing within a half-wavelength and assume Gaussian additive clutter. Extending virtual sensor-spacing beyond a half-wavelength could cause ambiguous DOA estimates.To settle these problems,this dissertation presents a new high-accurate unambiguous DOA estimation algorithm for CMIMO radar under SαS impulsive clutter. With transmit antennas spacing at a half-wavelength and receive antennas spacing beyond a half-wavelength,the proposed algorithm firstly utilizes the infinity-norm normalization to pre-process the received signal.Then,by exploiting the propagator method,the proposed algorithm uses the half-wavelength invafiance to yield high-variance but unambiguious DOA estimates and uses the larger invafiance to produce a set of cyclically ambiguous low-variance DOA estimates.Finally the high-variance estimates are served as a coarse reference to disambiguate the ambiguous low-variance DOA estimates.The new algorithm does not require the eigen-decomposition and spectral searching,hence,showing low computational complexity.4.Investigations on CRB for estimating DOA of CDT with CMIMO radarThis part studies the CRB for estimating DOA of CDT with CMIMO radar,and reveals the fundamental performances of CMIMO radar.Firstly,the general CRB expression is derived.Then,several special cases such as a CDT and a CDT with partial information are discussed in detail.Finally,the CRB properties for estimating a CDT are discussed.The results show that the CMIMO radar is superior to the traditional phased array radar in estimating the CDT parameters,and the CRB on estimating point target is superior to CDT for CMIMO radar.