Research On Models And Signal Processing For MIMO Radars

A Multiple-Input Multiple-Output (MIMO) radar transmits multiple probing sig-nals via multiple antennas,and receives the backscattered signals reflected from thetargets using multiple antennas.In contrast with the conventional phased array radar,the waveform diversity enables the MIMO radar with colocated antennas to have muchimproved capabilities including significantly improved parameter identifiability,muchenhanced flexibility for transmit beampattern design,and direct applicability of adap-tive array processing algorithms.The spatial diversity of the target's radar cross sectionobtained by the statistical MIMO radar enables improved detection and estimation per-formance.The problems of modeling of MIMO radar,the target detection,the parameterestimation,and optimal waveform designs are considered herein.The main contribution of this thesis are as follows.1.The recent results on signal models of MIMO radars are systematically sum-marized,including the MIMO radars with colocated antennas with waveform diversity,the statistical MIMO radars with transmit and receive diversity,the statistical MIMOradars with transmit diversity only,and the MIMO-STAP radars with colocated an-tennas.The statistical MIMO radar model with transmit diversity only is generalizedby introducing the target angular spread at the receive array.The equivalent model ofthe MIMO radars equipped with noncoherent transmit and receive colocated antennasubarrays is proposed.The multiple-target model of the MIMO-STAP system is alsoproposed.2.The point target model is used in the MIMO radars with colocated antennas,and hence the signals transmitted and received respectively via the transmit and re-ceive antenna arrays are fully coherent.In the statistical MIMO radars,the target ismodeled as extended with rich scattering,and the transmitted and received signals arecompletely decorrelated,respectively.This contribution introduces the partial spatialcoherence in the MIMO radars,i.e.,the signals illuminate on finite number of scat-terers of the target,while the backscattered signals are received via finite number of paths as well.Hence,the target exhibits finite angular spread with respect to both thetransmit and receive arrays.The parametric physical MIMO radar model is proposedto describe the partial spatial coherence.3.The MIMO radar with colocatd antennas,the statistical MIMO radars withtransmit and receive diversity,and the statistical MIMO radars with transmit diver-sity only,are constructed based on the physical scattering description.However,thesemodels cannot effectively capture the scattering characteristics,and it is not easy tohandle the problems which needs these information,such as the optimal waveform de-sign.Based on the parametric physical model of MIMO radars,exploiting the theory ofMIMO wireless communications and that of the bearnspace in array signal processing,the virtual representation model is proposed.In this model,the probing and reflectedsignals in different directions are respectively projected onto fixed Fourier directions,and the equivalent channel model is used to describe the scattering characteristics ofthe target.4.The canonical model of MIMO radars is further proposed,with the models ofMIMO radars with colocated antennas,the statistical MIMO radar,and that with partialspatial coherence as its special cases.The probing signals from different directionsare projected onto the transmit eigen directions,while the physical reflections of thetarget towards different directions are projected onto the receive eigen directions.Thescattering characteristics of different scatterers is described by a scattering matrix.Thismodel eases the universal optimal waveform design.5.For the statistical MIMO radar with transmit diversity only and angular spreadat the receive array,an efficient asymptotically optimal detection algorithm is proposedby use of the Toeplitz structure of the spatial coherence matrix,and the asymptoticequivalence of Toeplitz and circulant matrices,where the circulant matrix is diago-nalised by the discrete Fourier transform matrix.6.The direction finding algorithms,including Capon,APES,GLRT and transmitdiversity smoothing techniques are considered for the MIMO radars with colocatedantennas.The generalized MUSIC algorithm in conjunction with the transmissiondiversity smoothing technique are proposed in the parameter estimation problem of theMIMO radar with colocated antennas equipped with noncoherent transmit and receive colocated antenna subarrays.The spatial diversity at either the transmit or the receivecan be flexibly obtained by selecting certain transmit or receive subarrays.7.Several algorithms are proposed to deal with the problem of parameter estima-tion of the statistical MIMO radar with transmit diversity only.Under the quasistaticfading assumption,this contribution proposes two asymptotical maximum likelihoodmethods,and the CAPES algorithm to obtain the angle and fading parameters estima-tion,as well as the robust APES algorithm for fading estimation.Under the stochasticfading assumption,the covariance matching technique and the subspace algorithm areproposed for angle estimation.The application of blind signal processing in the param-eter estimation problem is discussed as well.The performance improvement of angleestimation with transmit diversity is validated through simulations.8.This contribution maximizes the conditional mutual information and minimizesminimum mean-square error respectively to obtain the optimal estimation waveformswith the canonical MIMO radar model.This result may be considered as a generalsolution to the statistical MIMO radar and MIMO radar with partial spatial coherence.The Kronecker structure matrix estimation technique is also applied to the problem ofoptimal waveform design.