Research On Moving Target Detection, Imaging And Parameter Estimation

The moving target detection (MTD) is one of the most important functions of radarsystem. The MTD performance is mainly dominated by the performances of cluttersuppression and target focusing. Clutter suppression can be realized effectivelyexploiting the degrees of freedom (DOF) of multiple antennas. Therefore, how toimprove the coherence of different channels and how to suppress the clutter effectivelybecome the most important subjects of MTD technique. Along with the development ofthe MTD research, the motion parameters estimation and the imaging of the movingtarget also become an important demand and a researching subject.This dissertation is mainly devoted to two subjects: one is how to improve the MTDperformance containing how to suppress the clutter and focus the targets effectively;one is how to estimate the target parameters effectively. The main research results are asfollows:1. The channel equalization and moving target detection and location formulti-channel SAR-GMTI,Improving the coherence of different channels is the key problem in MTD forairborne multi-channel SAR-GMTI system. Aiming at the problem, a novel channelequalization method is presented, and a technique for moving target detection (MTD)and location based on which is further proposed. This method first realizes circuitscalibration utilizing the leaking signals, then implements fast and slow time calibrationand compensates the channel differences via iteration after range compression, after thatrealizes MTD by clutter cancellation and imaging, finally obtains target parameters byestimating moving target Doppler shift. The channel equalization method needs nopriori knowledge such as the parameters of antenna and the plane’s movement, andavoids image registration and auto adaptive clutter cancellation method. Thus it’s ofless computation and easy to realize. The experiment result of real data proved that themethod can achieve a clutter cancellation ratio of more than20dB, and it has gooddetection and location performance.2. Multi-channel SAR-GMTI based on total least squares methodAiming at the poor clutter suppression performance induced by noise and poorclutter coherence in multi-channel SAR-GMTI, a method for SAR-GMTI based on totalleast squares is presented. The method adopts adjacent pixels around detection pixel tofit it under total least squares criterion, which can eliminate the influence of noise andimprove clutter coherence, thus lead to performance improvement on clutter suppression. Besides, the method is robust to error. Finally, simulation results and realdata processing are given to demonstrate the effectiveness of the method.3. The detection with high detection probability and the imaging with highaccuracy for multiple targets in multi-channel SARFor Multi-channel SAR-GMTI system, the movement of the targets leads to rangecell movement (RCM) and the shift of their Doppler rates. As the motion parameters ofthe targets are unknown, they are focused using the parameters of the stationary scene,which will induce their defocus and the degrade of their detection probability. Inaddition, the targets are extracted from SAR image and processed to obtain their motionparameters singly, which brings great computation complexity. To solve the problems, anew method to moving target imaging and detection with high detection probability ispresented. It first corrects RCM using Keystone transform and filter group whichcompensating phases of the target. Then LVD transform is performed to focusing thetargets. After detecting the moving targets in the LVD plane, the Doppler rates of themultiple targets are obtained. At last, targets imaging is realized in SAR image. Theprocessing of real data testifies that it can overcome RCM and the defocus induced bychirp rate shift, therefore improves the detection probability of targets efficiently.4. Clutter suppression algorithm for airborne bistatic sidelooking radarA simultaneous transmitter and receiver motion of the bistatic radar complicatesthe clutter spectra over range and makes the clutter spectra range-dependent [1].Range-dependent clutter makes space time adaptive processing (STAP) difficult toobtain independently and identically distributed (i.i.d.) secondary data used forcovariance matrix estimation, which will lead to the decline of the clutter suppressionperformance. To solve the problem, two clutter suppression algorithms for airbornebistatic sidelooking radar using MIMO or overlapped subarray alternate transmitting aredeveloped. In the methods, by incorporating the transmit degree of freedom (DOF), thebistatic clutter ridge turns to be a three-dimensional ridge. Moreover, the clutter ridgesof all range bins locate in the same plane of the three-dimensional space, while themoving targets don’t because of their radial velocity. For the clutter in the plane isrange-independent, it can be suppressed effectively by3-dimensional STAP.5. Fast algorithm for moving target space-time parameters estimationThe traditional method for moving target space-time parameters estimation needsrefined parameter searching, which leads to heavy computation burden. And itsestimation accuracy is related to the searching pace. To overcome the problem, two fastalgorithms for target space-time parameters estimation are developed: The maximum likelihood estimation based on polynomial rooting performs adaptive processing usingthe steering vectors of the three frequency channels neighbor to that of the target. In thisway, three adaptive weights and three response output are obtained. Then the space-timeparameters of the target are estimated by rooting a polynomial; the maximum likelihoodestimation based on quadratic polynomial approximation performs the quadraticpolynomial approximation of the spectrum of the frequency response using the outputof the three channels round which of the target. Then the space-time parameters of thetarget are estimated by calculating the extremum. The two methods effectively lower thecomputation complexity of the target space-time parameters estimation. Compared toparameter grid searching methods, their computation complexity is much decreasedwhile in the same accuracy.