Research On SAR/GMTI And Non-Sidelooking Linear Array STAP For Airborne Radar

Ground Moving Target Indication (GMTI) as a part of tactical reconnaissance is a necessaryfunction of the airborne surveillance radar for military application, and is a hot issue in the field ofradar signal processing. This thesis mainly investigates multi-channel Synthetic Aperture RadarGround Moving Target Indication (SAR/GMTI) technique and Space Time Adaptive Processing(STAP) technique for non-sidelooking Uniform Linear Array (ULA). The major work can besummarized as follows:1. Two-channel SAR/GMTI techniques based on eigen-decomposition of the covariancematrix are investigated. The previous studies turn out that the decomposition elements such as thesecond eigenvalue, the Along-Track Interferometric (ATI) phase and the similarity can be used asGMTI metrics. However, unfortunately, their GMTI performance is so low since those threemetrics only utilize the phase or amplitude information of the SAR image pair to detect movingtargets. To further enhance GMTI performance, two new composite metrics are introduced and areapplied to detect moving targets. One is called ellipse detector, jointing the similarity and the ATIphase information together and changing the two random variables into one random variable. Theother examines the statistic of the second eigenvalue and the ATI phase for ground moving targetindication. Based on this statistic, a Constant False Alarm Rate (CFAR) detector can be designedto solve the problem of GMTI. To detect slow moving targets more accurately, the secondeigenvalue and the ATI phase pre-thresholds are implemented before this CFAR detector. Finally,the detection capability of the two proposed methods is demonstrated by numerical experimentson simulated data and measured SAR data.2. Channel blind equalization techniques for multi-channel SAR/GMTI system areresearched. First, the echo model for multi-channel SAR/GMTI is built up. Second, the principleof channel blind equalization algorithm based on Eigen-Decomposition of data covariance matrixis investigated. However, it turns out that this algorithm has two fatal disadvantages. One is that itsuffers from a slow convergence rate. The other is that the effectiveness of this algorithm isseriously influenced by the moving target signal in training samples. Third, to improve itsconvergence rate, reduced-dimension technique is used into this algorithm and a new channelblind equalization algorithm is proposed. Experimental results on measured SAR data demonstratethat the proposed algorithm shows a fast convergence rate and is able to calibrate channelmismatch with much less sample support. However, unfortunately, this new algorithm is the sameto old one, and its effectiveness is also influenced by the moving target signal in training samples.Fourth, in order to enhance the robustness of the algorithm to moving target signal, median estimate is applied to the proposed algorithm. Finally, the validity of modified algorithm isdemonstrated by measured SAR data.3. Multi-channel SAR ground moving target detection and radial velocity estimation areinvestigated. At first, a novel approach to moving target detection is proposed for Multi-channelSAR system. This approach utilizes Multistage Wiener Filter to suppress clutter. To improveGMTI performance in heterogeneous clutter environment, this new approach also combinesDiagonal Loading (DL) techniques and modified Adaptive Power Residual Non-HomogeneityDetector (APR-NHD). Numerical experiments on measured SAR data are presented todemonstrate the validity and advantage of this new algorithm. Subsequently, two methods toestimate moving target’s radial velocity are introduced. One is the maximum likelihood estimation.The other is to use spare signal reconstruction for moving target’s velocity estimation. Comparedwith the former, the latter improves estimation precision of moving target’s velocity estimation.Finally, the validity of two estimation methods is demonstrated by Numerical experiments onmeasured SAR data.4. STAP technique for non-sidelooking ULA is researched. First, space-time two-dimensionaldistribution of clutter spectrum in non-sidelooking ULA is analyzed. Second, two non-adaptivecompensation methods for clutter range dependence such as Doppler Warping (DW) method andAngle Doppler Compensation (ADC) method are introduced. Meanwhile, Numerical experimentson simulation data are presented to analyze and compare compensation performance of twomethods. Third, two adaptive compensation methods for clutter range dependence such asAdaptive Angle Doppler Compensation (A2DC) method and Fast Adaptive Angle DopplerCompensation (FA2DC) method are introduced. FA2DC method is a new adaptive compensationmethod. In order to address the computational burden of A2DC method, FA2DC method insertsblock processing and Projection Approximation Subspace Tracking (PAST) technique into A2DCmethod. Finally, Numerical experiments on simulation data are presented to analyze and comparecompensation performance of two methods for clutter range dependence.